Taxonomy, Phylogeny and Diversity of the Extinct Lesser Antillean Rice Rats (Sigmodontinae: Oryzomyini), with Description of a New Genus and Species

See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/220025547

Taxonomy, phylogeny and diversity of the extinct Lesser Antillean rice rats (Sigmodontinae: Oryzomyini), with description of a new genus and species

Article in Zoological Journal of the Linnean Society · December 2010 DOI: 10.1111/j.1096-3642.2009.00628.x

CITATIONS READS 19 154

4 authors, including:

Samuel Turvey Marcelo Weksler Zoological Society of London Museu Nacional / UFRJ

83 PUBLICATIONS 1,612 CITATIONS 43 PUBLICATIONS 1,421 CITATIONS

SEE PROFILE SEE PROFILE

All in-text references underlined in blue are linked to publications on ResearchGate, Available from: Marcelo Weksler letting you access and read them immediately. Retrieved on: 29 August 2016 Zoological Journal of the Linnean Society, 2010, 160, 748–772. With 11 ﬁgures

Taxonomy, phylogeny, and diversity of the extinct Lesser Antillean rice rats (Sigmodontinae: Oryzomyini), with description of a new genus and species

SAMUEL T. TURVEY1*, MARCELO WEKSLER2, ELAINE L. MORRIS3 and MARK NOKKERT4

1Institute of Zoology, Zoological Society of London, Regent’s Park, London NW1 4RY, UK 2American Museum of Natural History (Division of Paleontology and Vertebrate Zoology), Central Park West at 79th Street, New York, NY 10024, USA 3Centre for Applied Archaeological Analyses, School of Humanities (Archaeology), University of Southampton, Avenue Campus, Southampton SO17 1BJ, UK 41, The Old Rectory, Fleggburgh Road, Rollesby, Norfolk NR29 5AJ, UK

Received 15 May 2009; accepted for publication 1 September 2009

Rice rats (Sigmodontinae: Oryzomyini) are abundant in the Late Quaternary fossil record and in Holocene pre-Columbian archaeological middens across the Lesser Antilles. All of these rice rats are now extinct, and their regional diversity and systematics remain extremely poorly understood. We redescribe all of the region’s rice rat taxa known from adequate diagnostic material (Megalomys desmarestii, Megalomys luciae, and Oligoryzomys victus), and describe a new genus and species, Pennatomys nivalis gen. et sp. nov., from archaeological sites on St. Eustatius, St. Kitts, and Nevis, which formed a single larger island during Quaternary low sea-level stands. Cladistic analysis supports the inclusion of O. victus within Oligoryzomys, and identiﬁes Megalomys as a sister group of the large-bodied genera Sigmodontomys or Sigmodontomys + Nectomys, suggesting that large body size in Megalomys represents phyletic gigantism rather than ‘island gigantism’. Megalomys and Pennatomys belong to an oryzomyine clade that has undergone remarkable radiation throughout the oceanic and continental-shelf islands of the Neotropical region, but these genera do not represent a monophyletic group within the Nectomys subclade, indicating multiple over-water colonization events of the Lesser Antillean island chain. Although Lesser Antillean rice rats were heavily exploited by prehistoric Amerindians, it is likely that most or all of these taxa survived until European arrival in the region.

ADDITIONAL KEYWORDS: extinct mammal – Megalomys – Nevis – Pennatomys – St. Eustatius – St. Kitts – West Indies – zooarchaeology.

INTRODUCTION nary fossil record and Holocene pre-Columbian archaeological sites throughout the main Lesser Anti- The Neotropical sigmodontine rodent tribe Ory- llean island chain (the Windward and Leeward zomyini (the rice rats) is one of the most species-rich Islands) (Steadman et al., 1984a; Pregill et al., 1994; New World mammal clades, with 28 extant genera Wing, 2001a, b; Newsom & Wing, 2004) (Fig. 1). Rice and approximately 115 species currently recognized rats probably occurred on all of the Lesser Antillean (Weksler, 2006; Weksler et al., 2006). Oryzomyines islands from Grenada northwards to the Anegada are also abundantly represented in the Late Quater- Passage, and the lack of records from a few islands in this chain (notably Dominica) almost certainly repre- *Corresponding author. E-mail: [email protected] sents a sampling artefact rather than a genuine

Anguilla St. Martin St. Barthelemy Saba Barbuda St. Eustatius (Megalomys audreyae) St. Kitts Nevis Antigua Montserrat

La Desirade Guadeloupe Marie Galante

Dominica

Martinique (Megalomys desmarestii)

St. Lucia (Megalomys luciae)

St. Vincent (Oligoryzomys victus) Barbados

Grenadines

Grenada

Figure 1. Map of the Windward and Leeward Islands of the Lesser Antilles, showing the distribution of extinct rice rats in Holocene archaeological sites (black stars) and Late Quaternary palaeontological sites (open circles), and islands from which rice rat species have been formally described. The modern-day sea level and 200 m isobath are both indicated, to show locations of shallow submarine banks that would have been exposed above water during Late Quaternary low sea-level stands, and which may have shared conspeciﬁc rice rat populations. Data from Pregill et al. (1994), Crock (2000), Newsom & Wing (2004), and LeFebvre (2007).

the past century, making wider systematic comparisons difficult (Friant et al., 1940; Friant, 1941; Stead- man & Ray, 1982), and no new rice rat species have been described from the Windward or Leeward Islands for over 80 years, despite the abundant material available in archaeological and palaeontological collections. Different authors have proposed various hypothetical intra-island (Steadman et al., 1984b; Lippold, 1991) and inter-island (Woods, 1989; Alcover et al., 1998; Turvey, 2009) species groupings for the Lesser Antillean rice rat radiation, but these remain impossible to evaluate without a better understanding of the extensive collections of currently undescribed rice rat material. Because rice rats comprised an important component of Amerindian diets, species concepts for extinct rice rats are further confused by the substantial evidence for prehistoric translocation Figure 2. Martinique giant rice rat (‘Le rat pilori, Mus pilorides’), possibly drawn from life, from Gervais of a wide range of land mammals and other verte- (1854: 411). brates between different Caribbean islands by Amer- indians (Miller, 1929, 1930; Newsom & Wing, 2004; absence from Late Quaternary faunas (Evans, 1968; Díaz-Franco & Jiménez Vázquez, 2008; Olson & Maíz Pregill et al., 1994). López, 2008). New taxonomic studies of West Indian All of the rice rats from the Windward and Leeward rice rats urgently need to be carried out, not only to Islands have become extinct since human arrival in clarify the diversity, systematics, ecology, and evolu- the insular Caribbean. Although they have frequently tion of this largely ‘forgotten’ group of mammals, but been mentioned in reviews or popular accounts of also to understand the dynamics and magnitude of historical-era mammal extinctions (e.g. Allen, 1942; late Holocene mammal extinctions at a global level. Balouet & Alibert, 1990; Alcover et al., 1998; MacPhee Undescribed rice rat remains, typically referred to & Flemming, 1999; Flannery & Schouten, 2001; ‘Oryzomys’ or ‘Oryzomyini’ in the archaeological lit- Turvey, 2009), they have so far been studied mainly erature, have been recorded from late Holocene by zooarchaeologists interested in Amerindian subsis- midden deposits on St. Eustatius, St. Kitts, and tence strategies and exploitation of biological Nevis, three islands located close together on the resources (e.g. Gullick, 1980; Jones, 1985; Nokkert, shallow submerged St. Kitts Bank in the Inner Vol- 1999; Newsom & Wing, 2004; Kozuch & Wing, 2006), canic Arc of the Leeward Islands (Fig. 1). Branch and regional species diversity and systematics remain (1907) ﬁrst noted that bones of a small mammal were extremely poorly understood. The only formally extremely plentiful in a midden on St. Kitts that was described rice rat species from the region known from exposed by a road cut, and rice rat remains are also adequate diagnostic material are the giant rice rats or abundant on this island in the post-Saladoid Sugar ‘pilories’ from Martinique and St. Lucia, Megalomys Factory/Sugar Factory Pier (AD 700–1000), Bloody desmarestii (Fischer, 1829) and Megalomys luciae Point (AD 660–1115), and Cayon (dates unknown) (Forsyth Major, 1901), and a pygmy rice rat from St. archaeological sites (Hoffman, 1973; Wing, 1973; Vincent, Oryzomys victus Thomas, 1898, which is now Wing & Scudder, 1980; Pregill et al., 1994; J.E. Robb, usually placed in the mainland Neotropical genus pers. comm.). They have also been reported on St. Oligoryzomys (e.g. Tate, 1932b; Hershkovitz, 1969; Eustatius from the Saladoid/post-Saladoid Golden Carleton & Musser, 1989; Musser & Carleton, 1993, Rock archaeological site (80 BC–AD 980) (van der 2005; MacPhee & Flemming, 1999). All three species Klift, 1992; Schinkel, 1992), and on Nevis from the were described from complete specimens collected Archaic-age site of Hichmans’ Shell Heap (790–520 during the 19th century (Fig. 2), and are thought to BC), the Saladoid site of Hichmans (100 BC–AD 600), have become extinct by the start of the 20th century and the post-Saladoid sites of Indian Castle (AD 650– (Allen, 1942; IUCN, 2008). A fourth extinct rice rat 880), Sulphur Ghaut (AD 900–1200), and Coconut species from the Leeward Islands, Megalomys Walk (dates unknown) by Wing (2001a, b), Newsom & audreyae Hopwood, 1926, was erected to describe Wing (2004), and Kozuch & Wing (2006). limited Late Quaternary fossil material from Large samples of zooarchaeological rice rat speci- Barbuda. mens from all three islands on the St. Kitts Bank, These four species have been the subject of very differing markedly in body size and craniodental mor- little morphological or taxonomic investigation over phology from Megalomys, were investigated, together

© 2010 The Linnean Society of London, Zoological Journal of the Linnean Society, 2010, 160, 748–772 EXTINCT WEST INDIAN RICE RATS 751 with existing historical specimens and/or zooarchaeo- wild-card taxa. A nexus ﬁle of the morphological and logical or palaeontological material of all previously molecular matrices is available in Appendix S1. described extinct rice rat species from the Windward The heuristic search algorithm implemented by and Leeward Islands. This forms the basis for a PAUP* 4.0b10 (Swofford, 2001) was used in all par- redescription of the region’s known rice rat taxa, a simony analyses. Each heuristic search employed description of a new genus and species of extinct 1000 replicates of random taxon addition with tree Holocene rice rat from the St. Kitts Bank, and the bisection–reconnection (TBR) branch swapping. Only cladistic analysis of these rice rats based on morpho- clades with at least one unambiguous synapomorphy logical character data to investigate their phyloge- (i.e. present in both ACCTRAN and DELTRAN char- netic relationships within the Oryzomyini and trends acter reconstructions; Wilkinson, 1995) were retained in insular evolution across the group. (commands PSET COLLAPSE = MIN; FILTER BEST in PAUP*). Nodal support was inferred via jackknife resampling and Bremer support (BS) values. Jack- MATERIAL AND METHODS knife values (Farris et al., 1996) for the parsimony MORPHOLOGICAL ANALYSIS analysis were calculated using 1000 pseudoreplicates, Dental terminology follows Reig (1987). Measure- with heuristic searches employed within each repli- ments were made using dial calipers, accurate to the cate (ten random addition replicates, TBR branch nearest 0.1 mm. Repositories of described or cited swapping, with no more than 20 trees saved per specimens are: Mammalogy and Palaeontology collec- replicate). Bremer support values (Bremer, 1988, tions, The Natural History Museum, London, UK 1994) were calculated with the help of TreeRot 3 (NHM); Environmental Archaeology Program, Florida (Sorenson & Franzosa, 2007) and PAUP*. Differences Museum of Natural History, Gainesville, FL, USA in tree length between the most-parsimonious uncon- (FLMNH-EAP). Examined material is listed under strained trees and trees constrained for alternative the systematic summary for each taxon. hypotheses of Megalomys relationships were tested using the Kishino–Hasegawa test (Kishino & Hase- gawa, 1989). CHARACTER SCORING AND PHYLOGENETIC ANALYSIS West Indian rice rats were scored for the characters RESULTS employed by Weksler (2006) in previous phylogenetic analyses of oryzomyines. Morphological characters SYSTEMATICS were combined with the molecular data available for ORDER RODENTIA BOWDITCH, 1821 extant oryzomyines (Irbp nuclear gene sequence; see SUPERFAMILY MUROIDEA ILLIGER, 1811 Weksler, 2003 and 2006 for details), and subjected to cladistic parsimony analyses. Changes in taxonomic FAMILY CRICETIDAE FISCHER, 1817 scope and in certain morphological characters from SUBFAMILY SIGMODONTINAE WAGNER, 1843 that of Weksler (2006) were made as described else- TRIBE ORYZOMYINI VORONTZOV, 1959 where (Weksler et al., 2006; McCain et al., 2007; Voss GENUS MEGALOMYS TROUESSART, 1881 & Weksler, 2009). All characters were equally weighted. The Irbp sequence characters were treated (TYPE SPECIES MUS PILORIDES SAY, 1822) as unordered, but some morphological characters Examined material: Megalomys desmarestii: NHM were ordered as described in Weksler (2006). Charac- 55.12.24.201 (skin and skull); NHM 50.11.30.5 ters with intraspeciﬁc variation were treated using (skin), and NHM 50.11.30.6 (skull and postcranial the ‘polymorphic’ coding of Wiens (1995; ‘composite’ skeleton) – these represent the same individual. coding of Weksler, 2006). Morphological characters, Megalomys luciae: NHM 53.12.16.2 (holotype; including state descriptions and illustrations, can be damaged skin and skull). viewed at the Morphobank repository at http:// Megalomys audreyae: NHM M7406a (holotype; left morphobank.org. mandibular ramus lacking m1); NHM M7406b We performed maximum parsimony phylogenetic (upper incisor). analyses on the combined morphological and molecular matrix, and on the morphological matrix only. In Diagnosis: Megalomys can be differentiated from all the combined analyses, nine taxa (including all other sigmodontine rodents by the combination of its extinct taxa) lack the Irbp sequence data, resulting in extreme body size, oryzomyine synapomorphies (e.g. many missing entries in the data matrix. This caused long palate, absence of both suspensory process of decreased nodal support values (see below), but we squamosal and alisphenoid struts), and several could not detect any other potential effect of extensive integumental, cranial, and dental traits, including missing data, such as a lack of resolution arising from nine synapomorphies (see below). Important charac-

Table 1. Measurement data for Megalomys desmarestii, Megalomys luciae, and Oligoryzomys victus (mm)

M. desmarestii M. desmarestii M. luciae O. victus Measurement (NHM 50.11.30.5-6) (NHM 55.12.24.201) (NHM 53.12.16.2) (NHM 97.12.26.1)

Greatest skull length 55.2 – – – Condylo-incisive length 53.6 – 47.5 – Breadth of occipital condyles 9.6 – 9.8 – Length of diastema 17.8 16.9 15.6 7.8 Palatal bridge 14.0 12.8 13.0 6.2 Length of incisive foramen 9.4 8.3 7.8 5.4 Breadth of incisive foramen 3.3 3.1 2.7 2.1 Length of maxillary molars 8.2 9.4 7.7 4.5 Breadth of ﬁrst maxillary molar 2.5 2.6 2.5 1.4 External alveolar breadth 10.7 10.5 10.4 5.6 Cranial height 15.7 14.6 13.7 9.3 Rostrum length 21.0 – – 11.2 Rostrum breadth 11.4 10.6 9.0 6.1 Least interorbital breadth 9.1 9.7 9.4 5.1 Orbital length 17.4 17.5 15.6 11.0 Zygomatic breadth 30.6 – 26.9 16.7 Breadth of braincase 16.9 – 15.1 11.4 Breadth of zygomatic plate 7.8 6.1 6.1 3.9 Head–body length 257 312 261 94 Tail length 254 278 145* 121 Hindfoot length (with claw) 45 59 50 25 Length of internal side of ear 18 20 18 14

*The tail of NHM 53.12.16.2 is incomplete.

ters include: pes without well-developed natatory base, but there are distinct paler areas under the chin fringes and interdigital webs, and with interdigital and throat and under the lower thorax, the two pads large and fleshy; robust skull with interorbital regions being separated by a streak of darker fur region convergent anteriorly, with well-developed (Fig. 3). Soft fur, subauricular patches absent. Pinnae supraorbital crests and squared braincase; zygomatic small, not reaching eye when laid forward. Mystacial plate lacks an anterodorsal spinous process; small and superciliary vibrissae not extending posteriorly incisive foramina, not extending posteriorly between beyond pinnae when laid back. Manual claws small M1 alveoli; large posterolateral palatal pits recessed and unkeeled. Hindfoot without well-developed nata- in deep fossae; stapedial foramen and posterior tory fringes and interdigital webs; tufts of ungual opening of alisphenoid canal small, and squamosal– hairs at bases of claws on dII–dV vestigial or absent; alisphenoid groove and sphenofrontal foramen plantar surface covered with distinct squamae distal absent. Well-developed capsular process of lower to thenar pad; hypothenar pad absent or vestigial in incisor alveolus; and superior and inferior masseteric M. luciae, present but reduced in M. desmarestii; ridges joined anteriorly as an open chevron. Molars interdigital pads large and fleshy, pads 1 and 4 con- are bunodont, with M1 anterocone divided by a tiguous with 2 and 3; claw of digit I (dI) extending shallow anteromedian flexus; accessory rootlets are just beyond first interphalangeal joint of dII; claw of present on M1 and m1; m2 and m3 have three roots. dV extending to middle of second phalange of dIV (Fig. 4). Tail about same length as head and body, Description (based on M. desmarestii and M. luciae sparsely haired and covered with conspicuous epider- only): Very large rats, as large or larger than any mal scales; without tuft of long terminal hairs, and extant oryzomyines (Table 1). Dorsal pelage dark, unicoloured (dark). Mammary complement unknown. blackish brown (almost black) (tan brown in one Skull with stout rostrum flanked by moderate or specimen in which the colour might have faded); deep zygomatic notches; interorbital region conver- ventral pelage abruptly paler in M. desmarestii, all gent anteriorly with well-developed supraorbital white, except for a few darker guard hairs; in M. crests (symmetrically constricted in an older adult luciae, general ventral colour is dark, with plumbeous specimen of M. desmarestii); braincase squared, with

Figure 3. Dorsal, lateral, and ventral views of Megalomys skins. Top, Megalomys desmarestii (NHM 55.12.24.201); bottom, Megalomys luciae (NHM 53.12.16.2). Scale bar: 5 cm. well-developed temporal crests; lambdoidal and mesopterygoid fossa penetrating anteriorly between nuchal crests well developed in older adults. Posterior maxillae; bony roof of mesopterygoid fossa completely margin of zygomatic plate dorsal to M1 alveolus; ossiﬁed in an older adult specimen of M. desmarestii, zygomatic plate lacks an anterodorsal spinous perforated by small (slit-like) sphenopalatine vacu- process. Jugal present and small (the maxillary and ities in a younger specimen of M. desmarestii and in squamosal zygomatic processes overlap). Nasal bones M. luciae. Alisphenoid strut absent (buccinator– with acutely angled posterior nasal margins; extend- masticatory foramen and accessory foramen ovale ing posteriorly behind lacrimals; lacrimals sutured conﬂuent). Alisphenoid canal with large anterior mainly to maxillary. Posterior wall of the orbit with opening. Stapedial foramen and posterior opening of postorbital ridge (faint in M. luciae). Frontosquamo- alisphenoid canal small; squamosal–alisphenoid sal suture anterior to frontoparietal suture. Parietals groove and sphenofrontal foramen absent; secondary with broad lateral expansions. Incisive foramina anastomosis of internal carotid crosses dorsal surface small, not extending posteriorly between M1 alveoli, of pterygoid plate (= carotid circulatory pattern 3 of spindle-shaped. Bony palate between the molar rows Voss, 1988). Posterior suspensory process of the is smooth or weakly sculpted. Posterolateral palatal squamosal absent. Postglenoid foramen small and pits large, complex, and recessed in deep fossae; dorsoventrally compressed, covered by tegmen

m1 anteroconid without an anteromedian ﬂexid; anterolabial cingulum present on all lower molars; ectolophid absent on m1 and m2; anterolophid present on m2 and m3; mesolophid distinct on unworn m1 and m2; m2 hypoﬂexid short; posteroflexid present on m3 (Fig. 7A). Accessory lingual and labial roots of m1 present; m2 and m3 each with two small anterior roots and one large posterior root. First rib with double articulation. Humerus without entepicondylar and supratrochlear foramina. Twelve ribs (false thirteenth rib present in older adult specimen). Fifth lumbar (seventeenth thoracicolum- bar) vertebra with well-developed anapophysis. Hemal arch between second and third caudal verte- brae, with posterior spinous process. Conditions of soft anatomy (stomach, liver, and male accessory reproductive glands) and glans penis characters are unknown.

Remarks: The complex early taxonomic history of Megalomys was described by Tate (1932a). Although historical museum specimens of M. desmarestii and M. luciae have permitted redescription of the genus and comparison of Megalomys with other extinct rice Figure 4. Plantar view of left hindfoot of Megalomys des- rats from the Windward and Leeward Islands, the marestii (NHM 50.11.30.6). Scale bar: 1 cm. status of other species that have been assigned to the genus requires further clarification through future research. The limited type material of M. audreyae (Fig. 7B) is difficult to evaluate, and the genus-level tympani; subsquamosal fenestra vestigial or absent. identity of this species remains unclear. Abundant Periotic exposed posteromedially, between ectotym- fossil and zooarchaeological rice rat material is panic and basioccipital, extending anteriorly to known from the Antigua–Barbuda Bank (Wing et al., carotid canal; mastoid unfenestrated. Capsular 1968; Steadman et al., 1984a; Jones, 1985; Pregill process of lower incisor alveolus well developed in et al., 1994; MacPhee & Flemming, 1999; Steadman & adult specimens (smaller in M. luciae); superior and Hilgartner, 1999), and fossil material from Barbuda inferior masseteric ridges joined anteriorly as open has been named ‘Ekbletomys hypenemus’byRay chevron below m1 (slightly anterior in M. desmarestii) (1962); however, this taxon remains a nomen nudum (Fig. 5). because Ray’s work has never been formally Upper incisors with smoothly rounded enamel published. Although further material apparently bands. Maxillary tooth rows parallel. Molars bun- referable to M. audreyae has been reported and odont and with labial flexi enclosed by a cingulum; radiometrically dated by MacPhee & Flemming labial and lingual flexi of M1 and M2 meet at midline, (1999), the identity of rice rats from the Antigua– enamel overlaps. M1 anterocone divided into antero- Barbuda Bank cannot be determined in the absence labial and anterolingual conules by shallow antero- of further investigation of this material, which is median flexus; anteroloph well developed and fused beyond the scope of the present study. with anterostyle on labial cingulum, separated from Two other large-bodied extinct insular cricetid anterocone by persistent anteroflexus; protostyle rodents, Megalomys curioi Niethammer, 1964 from absent; mesolophs present on all upper molars; para- the Galápagos Islands and Megalomys curazensis cone connected by enamel bridge to anterior or middle Hooijer, 1959 from Curaçao, were also originally moiety of protocone; median mure connected to pro- referred to the genus. Megalomys curioi has since tocone. M2 protoflexus absent; mesoflexus present as been transferred to the new genus Megaoryzomys single internal fossette; paracone without an acces- (Lenglet & Coppois, 1979) and reinterpreted as a sory loph. M3 with posteroloph and developed hypof- thomasomyine (Steadman & Ray, 1982). The southern lexus (the latter remaining excavated with moderate Netherlands Antilles show little biogeographical to heavy wear) (Fig 6A, B). Accessory labial root of M1 affinity with the Windward and Leeward Islands, and present. have close faunal and floral similarities to mainland

Figure 5. Cranium and mandible of Megalomys. Left to right: Megalomys desmarestii (NHM 50.11.30.6) and Megalomys luciae (NHM 53.12.16.2). Scale bar: 1 cm.

South America (Hooijer, 1959, 1966, 1967; Trejo- Diagnosis: Oligoryzomys victus can be diagnosed by Torres & Ackerman, 2001; Vázquez-Miranda et al., its small size, absence of jugal, absence of cranial 2007; Voss & Weksler, 2009), and it has been hypoth- crests, presence of large stapedial foramen and pos- esized that M. curazensis is derived from a different terior opening of alisphenoid, absence of squamosal– mainland oryzomyine ancestor (McFarlane & Lund- alisphenoid groove and sphenofrontal foramen, berg, 2002). A redescription of M. curazensis will form presence of large sphenopalatine vacuities, presence the basis for a future study in this series, and it is not of divided anterocone on M1, presence of two meso- considered further herein. fossetti on M2, and presence of ectolophid on m1. This diagnosis, however, is similar to that for South Ameri- GENUS OLIGORYZOMYS BANGS, 1900 can species of Oligoryzomys, and differentiation between insular and continental species requires (TYPE SPECIES ORYZOMYS NAVUS BANGS, further study (see below). 1899; = HESPEROMYS FULVESCENS SAUSSURE, 1860) OLIGORYZOMYS VICTUS (THOMAS, 1898) Holotype: NHM 97.12.26.1 (skin and skull). Description: Very small rodent (Table 1) with brown/ ochre dorsal coloration, greyish white ventral pelage, Examined material: Holotype only. and soft fur without subauricular patches (Fig. 8).

Figure 6. Maxillary molars of West Indian rice rats. A, Megalomys desmarestii (NHM 55.12.24.201). B, Megalomys luciae (NHM 53.12.16.2). C, Oligoryzomys victus (NHM 97.12.26.1). Scale bar: 1 mm.

Hindfeet without well-developed natatory fringes and etals without lateral process, interparietal strap interdigital webs, and with tufts of ungual hairs at shaped. Incisive foramina with posterior border ante- bases of claws on dII–dV. Plantar surface densely rior to molar alveoli, and with parallel lateral covered with distinct squamae distal to thenar pad borders. Bony palate between molar rows smooth and (smooth heel); hypothenar pad large, interdigital pads long, mesopterygoid fossa not extending anteriorly small and fleshy (pads 1 and 4 displaced proximally between maxillae. Posterolateral palatal pits not relative to 2 and 3). Claw of dI extends to first half of recessed in fossae. Bony roof of mesopterygoid fossa first phalange of dII; claw of dV extends to first perforated by very large sphenopalatine vacuities. interphalangeal joint of dIV. Tail longer than com- The alisphenoid strut is absent (buccinator– bined length of head and body, sparsely haired, and masticatory foramen and accessory foramen ovale are covered with more or less conspicuous epidermal confluent), and the alisphenoid canal has a large scales; it lacks a long tuft of terminal hairs, and is anterior opening. Stapedial foramen and posterior weakly bicoloured (dark above, pale below). The opening of the alisphenoid canal large, squamosal– mammae are inguinal, abdominal, postaxial, and pec- alisphenoid groove and sphenofrontal foramen absent toral, and number eight. (= carotid circulatory pattern 2 of Voss, 1988). Poste- Skull with delicate rostrum flanked by moderate rior suspensory process of the squamosal absent, zygomatic notches; interorbital region symmetrically postglenoid foramen large and rounded, and subsqua- constricted (amphoral shape), with rounded, mosal fenestra patent but covered by translucid mem- unbeaded supraorbital margins; braincase without brane. Periotic exposed posteromedially between the temporal, lambdoidal, and nuchal crests. Zygomatic ectotympanic and the basioccipital, extending anteri- plate without anterodorsal spinous process, its poste- orly to carotid canal. Mastoid perforated by conspicu- rior margin lying anterior to the M1 alveolus. Jugal ous posterodorsal fenestra. In the mandible, the absent (maxillary and squamosal processes in capsular process of the lower incisor alveolus is well contact). Nasal bones with blunt posterior margins, developed, and the superior and inferior masseteric not extending posteriorly beyond lacrimals, which are ridges converge anteriorly as an open chevron below equally sutured to maxillary and frontal bones. Pos- m1 (Fig. 9). terior wall of the orbit smooth, frontosquamosal The upper incisors have smoothly rounded enamel suture colinear with the frontoparietal suture. Pari- bands, and the maxillary tooth rows are parallel. The

Figure 7. Mandibular molars of West Indian rice rats. A, Megalomys desmarestii (NHM 55.12.24.201). B, Megalomys audreyae (NHM M7406a) (reversed for comparison). C, Oligoryzomys victus (NHM 97.12.26.1). Scale bars: 1 mm. molars are bunodont, with labial flexi enclosed by a of large stapedial foramen and posterior opening of cingulum, and longitudinal flexi not overlapping. M1 alisphenoid, and absence of squamosal–alisphenoid anterocone divided into anterolabial and anterolin- groove and sphenofrontal foramen (carotid circulation gual conules by deep anteromedian flexus; anteroloph pattern 2 of Voss, 1988); and presence of large sphe- well developed, separated from anterocone by nopalatine vacuities. The last three characters are persistent anteroflexus; protostyle absent; mesolophs among the recovered synapomorphies for Oligory- present on all upper molars; paracone connected by zomys listed by Weksler (2006), and the present phy- enamel bridge to anterior moiety of protocone; median logenetic results (see below) provide the first cladistic mure connected to protocone. M2 protoflexus absent; corroboration of the placement of O. victus within mesoflexus present as two internal fossetti; paracone that genus. without an accessory loph. M3 with posteroloph and Oligoryzomys victus consistently differs from other large hypoflexus (Fig. 6C). m1 anteroconid without studied species of Oligoryzomys in certain characters, anteromedian flexid, but with anteromedian fossettid; such as the presence of two mesofossetti in M2, and anterolabial cingulum present on all lower molars; presence of ectolophid on m1. However, the alpha anterolophid present on m1, but absent on m2 and taxonomy of the genus Oligoryzomys is among the m3; ectolophid present on m1 and m2; mesolophid most poorly known within the Sigmodontinae. distinct on unworn m1 and m2; posteroflexid present Although a revision of the northern South American on m3 (Fig. 7C). and Central American species (probable relatives of O. victus) is beyond the scope of this study, further Remarks: This species shares several characters with morphological comparisons of O. victus with species of the other species of the genus Oligoryzomys, such as: continental Oligoryzomys are still necessary to estab- small size; absence of jugal; smooth and amphoral lish its status as a distinct species from mainland supraorbital margins; absence of cranial crests; taxa. It is interesting to note that O. victus has caudad-oriented foramen magnum; divided antero- apparently not been recorded from pre-Columbian cone on M1; presence of posteroloph on M3; presence archaeological sites on St. Vincent, in marked

Figure 8. Dorsal, lateral, and ventral views of Oligoryzomys victus skin (NHM 97.12.26.1). Scale bar: 2 cm. contrast to rice rat taxa on other Lesser Antillean Etymology: From the Latin ‘pennatus’, winged or islands. This raises the possibility that the species feathered, combined with ‘mys’, the standard suffix may represent a relatively recent human-assisted for mouse. In honour of Elizabeth S. Wing, to com- translocation to the Windward Islands from an memorate her extensive contributions to Caribbean unknown source population in mainland South zooarchaeology. America. However, it is important to note that O. victus is substantially smaller in body size than other Diagnosis: Oryzomyine rodents with short and blunt described Lesser Antillean rice rats, and so may not nasals; dual articulation of lacrimal with maxillary have been perceived as a valuable food item by pre- and frontal; slightly convergent anterior interorbital Columbian Amerindians. In addition, a few unde- region; subtle supraorbital crests; posterior margin of scribed small-bodied rice rat specimens that may incisive foramen terminating immediately posterior represent this species have been reported from Amer- to anterior margin of alveolus of M1; long bony palate indian middens on the neighbouring islands of Car- (mesopterygoid fossa does not extend anteriorly riacou and Grenada (Lippold, 1991; LeFebvre, 2007). between maxillary bones); well-developed capsular Further investigation of archaeological sites in the process in mandibular ramus; and single anterior southern Windward Islands, and detailed morphologi- masseteric crest. M1 with undivided anterocone and cal studies of mainland Oligoryzomys species, are well-developed mesoloph, with anterior protocone– required to test these alternative hypotheses regard- paracone crista; M2 without protoflexus, and mesof- ing the native distribution and status of O. victus. lexus with single internal fossette; M3 with developed mesoloph, but posteroloph absent or vestigial, and hypoflexus narrow and disappearing with moderate GENUS PENNATOMYS GEN. NOV. wear. m1 with enclosed anteromedian fossettids, but Type species: Pennatomys nivalis gen. et sp. nov. without anteromedian flexid; with small ectolophid

Figure 9. Cranium and mandible of Oligoryzomys victus (NHM 97.12.26.1). Scale bar: 5 mm.

and ectostylid. Mesolophid and mesostylid present, Differences from other genera: From the available connected to entoconid by lingual cingulum. All lower material, Pennatomys does not seem to possess any molars with anterolabial cingula. M1–M3 with three remarkable anatomical feature that is not also shared alveolar roots, m1 with four roots (including acces- with other oryzomyines. The combination of several sory), m2 with three roots, and m3 with two roots. traits, however, differentiates this taxon from all Additional morphological information is provided other oryzomyines. Pennatomys displays three syna- under the species description below. pomorphies for a ‘Nectomys’ subclade within clade D

Figure 10. Maxillary and mandibular tooth rows and cranial elements of Pennatomys nivalis gen. et sp. nov. A–C, maxillary tooth rows. A, FLMNH-EAP 05270001 (St. Eustatius). B, NHM M 82456 (Nevis) (reversed for comparison). C, FLMNH-EAP 05270002 (St. Eustatius). D–G, mandibular tooth rows. D, NHM M 82452 (holotype: Nevis). E, NHM M 82460 (Nevis). F, FLMNH-EAP 05270003 (St. Eustatius) (reversed for comparison). G, FLMNH-EAP 05270004 (St. Eustatius) (reversed for comparison). H–I, maxilla, FLMNH-EAP 05270002 (St. Eustatius). J, mandible, labial view, FLMNH-EAP 05270003 (St. Eustatius). K, paired frontals, FLMNH-EAP 05270005 (St. Eustatius). Scale bars: 1 mm. of the Oryzomyini (see Weksler, 2006): absence of Type locality: Hichmans (site GE-5 of Wilson, 2006), protoﬂexus and presence of single mesofossette on M2 Saladoid-era Amerindian archaeological site, 100 (Fig 10A, B), and single anterior masseteric crest BC–AD 600; test pit 14, spit 4/2, context 1363, sample (Fig. 10J). Pennatomys can be differentiated from all BS 5402 (see Crosby, 2003 for speciﬁc site details); other members of the Nectomys subclade, including St. George, Gingerlands, Nevis, Federation of St. Megalomys, by a combination of the following char- Kitts–Nevis. acters: dual articulation of the lacrimal with the maxillary and frontal (other taxa have major articulation with the maxillary), absence of accessory root Paratypes: Four hemimandibles (NHM M 82453, M on M1 (labial accessory root present in other taxa), 82454, M 82459, and M 82460) from test pit 14, spit and absence of posteroloph (posteroloph present in 4/2, context 1363, sample BS 5402; one maxilla (NHM other taxa). Additional character differentiation M 82455) from test pit 14, spit 4/1, context 1362, among the three described oryzomyine taxa from the sample BS 5344; one maxilla (NHM M 82456), and Lesser Antilles (Pennatomys, Megalomys, and O. one humerus and one femur (NHM M 82458), from victus) is shown in Table 2. test pit 14, spit 2/2, context 1362, sample BS 5319; paired frontals with associated left nasal (NHM M 82457) from test pit 14, spit 2/2, context 1362, sample PENNATOMYS NIVALIS GEN. ET SP. NOV. DS 5322; one premaxilla with associated incisor Holotype: NHM M 82452, right hemimandible with i1 (NHM M 82461) from test pit 14, spit 5, context 1363, and m1–3 (Fig. 10D). sample BS 5407; all from Hichmans (Fig 10B, E).

Table 2. Condition of morphological characters for West Indian rice rat taxa

Character trait Pennatomys Megalomys Oligoryzomys

Size Medium Large Small Pedal plantar pads Unknown Hypothenar absent or Hypothenar present, vestigial, digitals large digitals small Nasals posterior extension Even with Posterior to Even with maxillary–frontal– maxillary–frontal– maxillary–frontal– lacrimal suture lacrimal suture lacrimal suture Lacrimal articulation Equal maxillary and Primarily with maxillary Equal maxillary and frontal frontal Supraorbital shape Slightly convergent Convergent anteriorly Squared, smooth anteriorly with weak with well-developed margins crest crest Posterior margin of Slightly anterior to M1 Approximately even with Anterior to M1 alveolus zygomatic plate alveolus M1 alveolus Carotid circulation pattern Unknown III* II* Anterior section of Joined Separated Separated masseteric crests in ramus Labial accessory root on M1 Absent Present Unknown† Roots on m2 3 3 Unknown† Roots on m3 2 3 Unknown† Anteromedian ﬂexus on M1 Absent Present, shallow Present, deep Mesofossette on M2 Single Single Double Posteroloph on M3 Absent Present Present Anterolophid on m2 and m3 Present Present Absent

Megalomys characters are based on Megalomys luciae and Megalomys desmarestii only. *Pattern II, stapedial foramen present, sphenofrontal foramen absent; pattern III, stapedial foramen and sphenofrontal foramen absent. †Other species of Oligoryzomys do not have a labial accessory root on M1, and have only two roots on m2 and m3.

Examined specimens: In addition to the type series, far as nasals. Interorbital region slightly convergent additional dissociated craniodental and postcranial anteriorly, with subtle supraorbital crests. Postorbital rice rat skeletal material was also examined from the ridge apparently absent; frontosquamosal suture col- archaeological sites of Sugar Factory Pier, St. Kitts linear with frontoparietal suture, dorsal facet of (FLMNH-EAP 02510001-02510008; see Wing, 1973); frontal not in contact with squamosal. Zygomatic and Golden Rock, St. Eustatius (FLMNH-EAP plate broad, with anterodorsal margin smoothly 05270001–05270005; see van der Klift, 1992). rounded and conspicuously anterior to superior maxillary root of zygoma, and posterior margin situated Distribution: Known from late Holocene pre- slightly anterior to alveolus of M1. Posterior margin Columbian zooarchaeological sites on the islands of of incisive foramen terminates immediately posterior St. Eustatius, St. Kitts, and Nevis. to anterior margin of alveolus of M1. Bony palate long and relatively flat, with shallow lateral excavations; Etymology: From the Latin ‘nivalis’, snowy. In refer- mesopterygoid fossa does not extend anteriorly ence to ‘Nuestra Señora de las Nieves’, ‘Our Lady of between maxillary bones. the Snows’, the early Spanish name for Nevis, appar- Capsular process of lower incisor alveolus present ently so-called because of misidentification of the in the mandible, situated either ventral to the coro- clouds surrounding Mount Nevis. noid process or between the coronoid process and the condyle process. Superior and inferior masseteric Description: No complete or partial skulls known; ridges often conjoined anteriorly as single crest, cranial morphology based on dissociated elements which terminates ventral to anterior margin of m1. (Fig. 10H–K). Nasals short, terminating anterior to or Upper incisors with smoothly rounded enamel extending slightly further than the maxillary– bands. Maxillary tooth rows parallel. Molars bun- frontal–lacrimal suture, and with blunt posterior odont and brachyodont, with labial flexi enclosed by a margin. Premaxillaries extend posteriorly almost as cingulum; labial and lingual flexi meet at midline,

© 2010 The Linnean Society of London, Zoological Journal of the Linnean Society, 2010, 160, 748–772 762 S. T. TURVEY ET AL. enamel overlaps. All upper molars with three roots m1 with enclosed anteroflexid and anteromedian (accessory labial root of M1 absent). m1 with two fossettids crowded together, each with an anterior large roots and two central accessory roots, m2 with extension close to the crown midline; coalescing into a one large posterior and two smaller anterior roots, single transverse fossettid with minor wear. Antero- and m3 with one large anterior and one large poste- labial cingulum present, fused with anterolabial rior root (Fig 10A-G). surface of protoconid. Protoflexid narrowly com- M1 anterocone undivided and well developed (equal pressed labially, with apex bifurcated, and obliquely in length and width to protocone–paracone). Internal anterolinguad, almost reaching anterior fossettids; fold of procingulum circular, medially situated, and can become isolated internal island with moderate eliminated by heavy wear in older specimens. Anter- wear. Hypoflexid transverse broad, deep, and open, oloph reaching labial margin, separated from antero- extending halfway across crown. Small ectolophid and cone by short anteroflexus, which extends to internal ectostylid present. Metaflexid small, narrow, antero- fold and disappears with slight wear. Protostyle posteriorly elongated or obliquely anterolinguad; absent; protoflexus broad and deep, with large, gently becomes isolated island with slight wear, and absent squared apex. Paraflexus transversely oriented from with moderate wear. Metafossetid deep and open, labial wall, deflected posteriorly close to crown coalesced with deep and curved mesoflexid with slight midline, and extends along entire length of paracone. wear, to almost completely outline the metaconid. Mesoloph well developed; mesoflexus short, trans- Mesolophid and mesostylid present, connected to verse, with slightly expanded apex. Paracone con- entoconid by lingual cingulum. Small entoflexid only nected by enamel bridge to anterior or middle moiety present on unworn teeth. Posteroflexid and postero- of protocone; median mure connected to protocone. fossetid coalesce with slight wear, and outline Hypoflexus slightly deeper than protoflexus, with tri- entoconid. Posterofossetid deep, slightly expanded angular apex. Metaflexus deep, crescentic, extending anteriorly; apex deflected posteriorly. Posteroflexid over halfway across crown, almost reaching posterior relatively straight, obliquely anterolabial. Metaconid wall, and reaching hypoflexus. Posteroflexus small, and entoconid subquadrate; protoconid and hypoconid obliquely anterolingual notch at posterior margin of subtriangular. metacone. Very small internal fossette situated m2 with shallow protoflexid, occupying 30% of the between posteroflexus and posterolabial margin of crown width, and obliquely anterolinguad at 45°; metaflexus, sometimes discernible on worn teeth. defines shelf-like anterolabial cingulum. Hypoflexid M2 protoflexus absent; mesoflexus present as single broad and deep, but without apical bifurcation. internal fossette; paracone without accessory loph. Metaflexid small, becoming transverse island in Paraflexus slightly posterolinguad, extending halfway anterolingual corner of metaconid. Mesolophid across crown. Mesoflexus very reduced, defining small present, joined to entoconid by lingual cingulum. triangular mesoloph. Mesoflexus internal fossette Mesoflexid almost completely circumscribes meta- usually slightly anterolinguad and anteroposteriorly conid, very broad and deep, extending over half of the constricted medially. Hypoflexus very deep, sometimes crown width, and widening gradually away from with slightly rounded, expanded apex, and anteropos- lingual wall; apex coalesced with transversely ellipti- teriorly shorter than on M1. Metaflexus crescentic, cal first internal fold. Posteroflexid very deep, even deep and broad, extending well over halfway across with slight wear; extends to midline, becoming more crown. Posteroflexus very small and faint, apparently transverse with wear. Entoflexid slightly obliquely apically bifurcated. posterolabiad, with distinct short medial anterior M3 with developed mesoloph, but posteroloph extension, extends labially just beyond crown midline. absent (or vestigial; joined to metacone with little Anterolabial cingulum and protoflexid of m3 similar wear). Hypoflexus narrow and extremely small in to condition in m2, but with protoflexid slightly more specimens with little wear, disappearing with moder- transverse and shallow. Hypoflexid prominent, occu- ate to heavy wear. Paraflexus slightly posterolinguad, pying over half of the crown width, but anteroposte- broad and deep on unworn teeth, becoming greatly riorly shorter than in m2; gently curved, or with reduced by wear; can form separate small internal anterior margin curved and posterior margin straight fold adjacent to apex. Mesoflexus large, transverse; and obliquely anterolabial at 45°. Metaflexid small, can become isolated as an island. Paracone trans- transverse island in anterolingual corner of meta- verse, anteroposteriorly short or triangular; almost conid; disappears with slight wear. Mesoflexid long, isolated by paraflexus and mesoflexus. Metaflexus deep, occupying 70% of the crown width, and curving posterolinguad, anteriorly convex, extending 75% of around metaconid, with slight bifurcation at medial the distance across crown, often reaching hypoflexus apex. Entoconid very small and subtriangular. Entof- region; can coalesce distally with diminutive poster- lexid and posteroflexid distinct only on unworn teeth, oflexus to develop bifurcated apex. otherwise coalesced. Entoflexid 50–60% of the crown

Table 3. Tooth row measurements of Pennatomys nivalis gen. et sp. nov. populations from Nevis, St. Eustatius, and St. Kitts (mm)

Nevis St. Eustatius St. Kitts

Maxillary molar row (occlusal) 6.12 (5.6–6.4) 6.39 (6.1–6.7) 6.57 (6.4–6.7) SD = 0.249, n = 14 SD = 0.188, n = 14 SD = 0.137, n = 6 Mandibular molar row (occlusal) 6.61 (5.9–7.1) 6.65 (6.4–6.9) 6.84 (6.4–7.4) SD = 0.237, n = 41 SD = 0.141, n = 22 SD = 0.240, n = 13

Size range, standard deviation, and sample size given for each measurement. width, and 35–40% of the crown length; obliquely three populations, possibly associated with the ‘island posterolabiad at almost 45°; with anterior extension rule’ (terrestrial vertebrate body mass increases with close to lingual margin. Posteroflexid small and increasing island area; Burness et al., 2001), as St. transverse. Kitts (174.2 km2) is approximately twice the size of Nevis (92.5 km2). However, this pattern is not Remarks: Our description of a new genus for this straightforward, as St. Eustatius, the smallest of the taxon is based on the morphological distinctiveness of three islands (21 km2), contains rice rats that do not the new material, both in size (Tables 1, 3) and in differ significantly in body size from the St. Kitts qualitative craniodental characters (Table 2), and in population by either tooth-row measurement. As all the results of phylogenetic analyses. The corrobora- examined material was retrieved from archaeological tion of this hypothesis awaits the discovery of addi- middens, these observed size differences might also be tional material that would allow for a more complete explained by differential patterns of Amerindian phylogenetic assessment. exploitation of rice rats on different islands. This Although preliminary analysis of zooarchaeological alternative hypothesis is supported by the observa- material from Hichmans suggested that two rice rat tion that femora and humeri with completely fused size morphs might be found at the site (Nokkert, epiphyses were common, and maxillary and mandibu- 2002), more detailed morphological and morphometric lar dentition was moderately/severely worn in all rice investigation indicates that only a single rice rat rat specimens in the St. Kitts sample, whereas taxon is present (Table 3). The slight variation in although some mature individuals were also present body size of some specimens (e.g. NHM M in samples from the other islands (see above), fused 82459; = ‘Megalomys’ of Nokkert, 2002) is associated limb bones were rare or absent, and less than 50% of with increased age and extreme tooth wear. tooth rows exhibited strong wear in the St. Eustatius The islands of St. Eustatius, St. Kitts, and Nevis and Nevis samples, indicating that the St. Kitts share a submerged bank that would have formed a sample consisted of older rice rat individuals. Longer- single, larger island during low sea-level stands in term temporal patterns of body size change in rice rat Quaternary glacial periods (Pregill et al., 1994; populations apparently driven by Amerindian overex- Lambeck & Chappell, 2001). They can therefore be ploitation of larger individuals have also been identi- considered as a single biogeographic unit, with only fied on other neighbouring Lesser Antillean islands episodic barriers to gene flow between different popu- (Nokkert, 1999; see below). Because of these difficul- lations of a single species. No morphological qualita- ties in interpreting minor body size differences in tive character was observed to vary between the three otherwise morphologically uniform rice rats from island samples. However, minor but statistically sig- sites across the St. Kitts Bank, all three island popu- nificant differences in maxillary and mandibular lations are interpreted here as representing the same tooth-row length (interpreted as an indirect measure species. of overall body size) were observed between different Although P. nivalis gen. et sp. nov. is an abundant sampled island populations of Pennatomys from the component of Saladoid and post-Saladoid archaeologi- St. Kitts Bank. The Nevis sample is significantly cal sites on the St. Kitts Bank, there are no confirmed smaller in mandibular tooth-row length than the St. records of its survival in the European historical Kitts sample, and is significantly smaller in maxillary period. The Englishman George Percy, who stopped at tooth-row length than both the St. Kitts and St. Nevis around 1606, reported that his men went Eustatius samples (one-way ANOVA: maxillary tooth hunting on the island and ‘got great store of Conies’ row length, F = 11.14, P < 0.001; mandibular tooth (Wilson, 2006), but this record almost certainly refers row length, F = 5.18, P = 0.008; Table 3). This pattern to agoutis (Dasyprocta), which had been translocated may reflect genuine biological differences between the across the Windward and Leeward Islands from

© 2010 The Linnean Society of London, Zoological Journal of the Linnean Society, 2010, 160, 748–772 764 S. T. TURVEY ET AL. mainland South America by Amerindians in earlier only five steps longer, a difference that is statistically centuries (Newsom & Wing, 2004). Sir Henry Colt not significant (P > 0.05) in the Kishino–Hasegawa reported in 1631 that animals specifically identified test. as rats provided ‘good meat’ on St. Kitts (Harlow, Analysis of the morphological data set resulted in 1925), and in 1720 the Reverend William Smith five most-parsimonious trees (539 steps; CI = 0.26; claimed that ‘in Nevis some people do eat Rats, wrap- RI = 0.63), the strict consensus of which (tree not ping them up in Banano-leaves to bake them as it shown) again recovers the same overall phylogenetic were under warm Embers’ (Merrill, 1958). However, structure of a previous analysis of morphological black rats (Rattus rattus) have also been eaten in characters (Weksler, 2006) with the addition of the many parts of the West Indies during recent history newly scored West Indian rice rat taxa. Megalomys is (Merrill, 1958; Higman, 2008), and these historical recovered as a sister group of the clade containing records may conceivably also refer to exotic murids Nectomys and Sigmodontomys, whereas O. victus is rather than endemic rice rats. Intriguingly, there are recovered as the most basal scored member of the several reports of unusual-looking rats occurring on genus Oligoryzomys. Pennatomys is recovered as a Nevis into recent times, and being eaten by inhabit- sister group of the clade containing Amphinectomys, ants of the island until at least the 1930s (J. Johnson, Aegialomys, Nesoryzomys, Melanomys, and the Mega- Nevis Historical and Conservation Society, pers. lomys plus Nectomys–Sigmodontomys clade. Nodal comm.). support for most clades in the tree is low (as also recovered by Weksler, 2006), but is high for the M. luciae–M. desmarestii node (jackknife = 99%; BS = 6). The placement of O. victus as the most basal scored CLADISTIC ANALYSIS member of Oligoryzomys also received moderate We scored M. desmarestii for 81 (83%), M. luciae for support (jackknife = 62%; BS = 2), but nodes in the 71 (72%), O. victus for 73 (74%), and P. nivalis gen. et tree area of Pennatomys are uniformly low. Trees sp. nov. for 49 (50%) morphological characters. Mega- constrained for inclusion of the new material in Mega- lomys desmarestii and O. victus were scored for most lomys are six steps longer, a difference that is again external, cranial, skeletal, and dental characters, not significant in the Kishino–Hasegawa test. whereas M. luciae was scored only for dental and cranial characters, and P. nivalis gen. et sp. nov. was scored for dental, mandibular, and very few cranial DISCUSSION characters. By comparison, other oryzomyines, previously scored in Weksler (2006), ranged in complete- PHYLOGENETIC RESULTS ness from 32% (Amphinectomys) to 100% (two taxa). Previous proposals for relationships of Megalomys are Parsimony analysis of the concatenated matrix of rejected by our analyses. In the most comprehensive morphological and molecular (Irbp) data sets resulted comparative study of West Indian rice rat material in four trees (1245 steps; consistency index, CI = 0.37; available to date, the unpublished thesis of Ray retention index, RI = 0.64), the strict consensus of (1962), Megalomys was considered to be closely which recovers the same overall phylogenetic struc- related to Mindomys hammondi (then assigned to ture of a previous analysis of the oryzomyine data set Oryzomys; see Weksler et al., 2006 for the current (Weksler, 2006) with the addition of the newly scored oryzomyine classification). In fact, Ray (1962) West Indian taxa (Fig. 11). Oligoryzomys victus is included Mindomys hammondi as a member of the placed as the most basal scored member of the Oli- subgenus Megalomys within Oryzomys (see also Her- goryzomys clade, whereas Megalomys is recovered as shkovitz, 1970). However, this phylogenetic hypoth- the sister group of Sigmodontomys aphrastus. Pen- esis is falsified by our analyses. Mindomys is never natomys is recovered as the sister group of the clade recovered within clade D (see also Weksler, 2006), containing Aegialomys, Melanomys, Sigmodontomys, where Megalomys is robustly placed. Trees containing Nectomys, and Amphinectomys. Nodal support is high Mindomys and Megalomys as sister taxa are nine for Megalomys (jackknife = 99%; BS = 5), and lower steps longer than the most-parsimonious tree (the for Oligoryzomys including victus (jackknife = 56%; difference is not significant in Kishino–Hasegawa BS = 2), but the placement of Pennatomys is much tests, but both taxa lack molecular data). less secure (jackknife < 50%; BS = 1). There is a steep McFarlane & Lundberg (2002), in turn, rejected the decline in nodal support involving the region on the close relationship between Megalomys and Mindomys tree leading to Pennatomys, as expected from the hammondi on biogeographic grounds, and suggested fragmentary nature of the new material. West Indian that Megalomys was more closely related to an ory- rice rats are never recovered as a monophyletic unit: zomyine taxon inhabiting the Caribbean coast of trees constrained for Pennatomys and Megalomys are South America. Our results also do not corroborate

to outgroups 64 Scolomys ucayalensis 100 Zygodontomys brevicauda A 10 Zygodontomys cherriei 81 Euryoryzomys macconnelli 88 Euryoryzomys lamia 4 Euryoryzomys russatus Mindomys hammondi Oecomys catherinae 96 63 Oecomys bicolor 98 3 58 Oecomys trinitatis 5 96 Oecomys concolor 5 Oecomys mamorae B 61 Handleyomys intectus 73 Handleyomys alfaroi Handleyomys chapmani Handleyomys rostratus 98 Hylaeamys megacephalus 3 Hylaeamys yunganus 54 60 Transandinomys talamancae 98 Nephelomys albigularis 6 Nephelomys levipes 97 Neacomys spinosus 6 100 Neacomys minutus 5 Neacomys musseri 77 81 Microryzomys minutus 2 3 Oreoryzomys balneator 53 C Oligoryzomys victus 2 56 Oligoryzomys fulvescens 2 90 86 Oligoryzomys flavescens 2 2 Oligoryzomys fornesi 56 Oligoryzomys nigripes 1 Oligoryzomys stramineus Eremoryzomys polius Sooretamys angouya Cerradomys subflavus 96 Oryzomys couesi 63 4 Oryzomys palustris 96 Pseudoryzomys simplex 5 91 Holochilus brasiliensis 5 Lundomys molitor 99 Nesoryzomys narboroughi 3 Nesoryzomys swarthi D Pennatomys nivalis Aegialomys xanthaeolus 56 Melanomys caliginosus Sigmodontomys alfari 86 80 Nectomys squamipes 4 3 Amphinectomys savamis Nectomys apicalis Sigmodontomys aphrastus 99 Megalomys desmarestii 5 Megalomys luciae

Figure 11. Phylogenetic relationships of West Indian oryzomyines. Consensus cladogram of four most-parsimonious trees (tree length, L = 1245; consistency index, CI = 0.37; retention index, RI = 0.64) of combined molecular (Irbp) and morphological characters. Jackknife (> 50%) and Bremer (> 1) nodal support indices are shown above and below branches, respectively. Clades A, B, C, and D are the same as those recovered by Weksler (2006). Out-groups include Peromyscus maniculatus (Neotominae), Nyctomys sumichrasti (Tylomyinae), and Delomys sublineatus, Thomasomys baeops, and Wiedomys pyrrhorhinos (Sigmodontinae). Oryzomyine generic taxonomy follows that of Weksler et al. (2006). this scenario, as S. aphrastus, the recovered sister netic position of S. aphrastus is controversial taxon of Megalomys, is a species known from scat- (Weksler, 2006; McCain et al., 2007), and the present tered Transandean localities in Ecuador, Panama, hypothesis of Megalomys plus S. aphrastus needs and Costa Rica (McCain et al., 2007). The phyloge- corroboration from additional data. Nevertheless, the

© 2010 The Linnean Society of London, Zoological Journal of the Linnean Society, 2010, 160, 748–772 766 S. T. TURVEY ET AL. recovered close relationship of Megalomys to Necto- recorded in contemporary historical accounts about mys and related taxa (e.g. Sigmodontomys and Mela- the ecology of now-extinct rice rat species (Allen, nomys) has been suggested by several authors (e.g. 1942), but some clues are provided by the foot mor- Allen, 1912; Musser & Carleton, 2005). phology of Megalomys (Stein, 1988; Voss, 1988; The newly described genus Pennatomys is also Weksler, 2006). Megalomys possesses a markedly dis- recovered within clade D in our analyses, but inter- tinct plantar morphology from closely related ory- estingly Pennatomys and Megalomys are not recov- zomyines, and does not display well-developed ered as sister groups within this clade. This suggests natatory fringes or interdigital membranes, indicat- that the Lesser Antillean island chain was colonized ing that representatives of the genus were terrestrial by multiple over-water dispersal events by represen- rather than semi-aquatic. This may reflect the rela- tatives of clade D. However, support for the phyloge- tive lack of large rivers and streams on islands in the netic position of Pennatomys within clade D is low, Lesser Antilles. Instead, Megalomys possesses other and a more detailed understanding of the morphologi- morphological features, such as a long tail and well- cal character state combinations displayed by other developed large, fleshy digital pads, which are found undescribed rice rat taxa from the Windward and in arboreal or semi-arboreal (scansorial) rodents (e.g. Leeward Islands is required to investigate this phy- Oecomys among oryzomyines). Several other extinct logenetic hypothesis further. medium- to large-bodied rodents from the insular Our analyses corroborate the inclusion of O. victus Caribbean are also believed to have been arboreal or in the genus Oligoryzomys, as proposed by several scansorial, on the basis of skeletal adaptations, and researchers of oryzomyine systematics. Its basal all of the surviving West Indian rodents show varying position in the genus indicates its distinctiveness degrees of arboreality (Turvey et al., 2006). from the other scored species, but further alpha Phylogenetic analysis of the extinct West Indian taxonomy studies are necessary to corroborate its rice rats also provides new insights into the nature of status in relation to continental Oligoryzomys species body size evolution in these insular oryzomyines, and from Venezuela and Panama (e.g. Oligoryzomys challenges previously held assumptions about the fulvescens). evolution of Megalomys. The body size of Megalomys is among the largest of all sigmodontines, and this EVOLUTION, BIOGEOGRAPHY, AND ECOLOGY genus has been interpreted by several authors as a Megalomys and Pennatomys belong to a clade of ory- classic example of autapomorphic gigantism or ‘island zomyines (clade D of Weksler, 2006) that has under- gigantism’ derived from a small-bodied ancestor (e.g. gone a remarkable radiation throughout the oceanic Steadman & Ray, 1982; McFarlane & Lundberg, and continental shelf islands of the Neotropical 2002). The closest relatives of Megalomys, however, region, including taxa in the Galápagos (Aegialomys are also some of the largest living oryzomyines and Nesoryzomys; Patton & Hafner, 1983), Fernando (Weksler, 2006). Sigmodontomys aphrastus, the sister de Noronha (Noronhomys; Carleton & Olson, 1999), taxon of Megalomys, is a medium-sized rodent (head– Curaçao (Oryzomys; McFarlane & Debrot, 2001; Voss body length = 152 mm; McCain et al., 2007), but some & Weksler, 2009), Trinidad (Nectomys; Hershkovitz, specimens of Nectomys are recorded as having a 1944), and Jamaica (Oryzomys; Thomas, 1898; head–body length of 254 mm (Weksler, 2006). This Musser & Carleton, 2005), as well as the Windward suggests that the alternative hypothesis of phyletic and Leeward Islands. Several taxa within clade D gigantism may also explain the large body size of (e.g. Lundomys, Oryzomys, Holochilus, Nectomys, Megalomys, and emphasizes that phylogenetic analy- Pseudoryzomys,andSigmodontomys) have markedly ses are required before making assumptions about semi-aquatic adaptations, such as the development of body size evolution in insular taxa. natatory fringes and interdigital membranes, and are associated with marshes, rivers, or streams (Stein, 1988; Santori et al., 2008). This ecology may have TIMING AND CAUSATION OF THE ‘FORGOTTEN’ RICE predisposed members of the clade for over-water RAT MASS EXTINCTION flotsam dispersal from mainland South America to The only native Quaternary non-oryzomyine land the insular Caribbean, as they probably have an mammal known from the Windward and Leeward increased likelihood of becoming associated with rafts Islands, the giant heptaxodontine rodent Amblyrhiza of vegetation that could get carried into ocean cur- inundata, is believed to have died out as a result of rents, notably the south-east–north-west surface catastrophic habitat loss associated with Late Pleis- current flow of the North Atlantic Gyre (see Hedges, tocene sea level changes, which reduced island area 2006). on the Anguilla Bank before humans colonized the In contrast, West Indian rice rats do not appear to West Indies (McFarlane et al., 1998). Conversely, the have been semi-aquatic. Minimal information is region’s rice rat fauna survived into the period of

© 2010 The Linnean Society of London, Zoological Journal of the Linnean Society, 2010, 160, 748–772 EXTINCT WEST INDIAN RICE RATS 767 human colonization of the West Indies during the late extinctions remains impossible in the absence of his- Holocene. Megalomys desmarestii, M. luciae, O. torical last-occurrence dates for most taxa, to indicate victus, and possibly also the Nevis and St. Kitts how long they persisted after European arrival. populations of P. nivalis gen. et sp. nov. and the However, although these rice rats evolved in the undescribed rice rat from Barbados are all known presence of avian predators, notably a radiation of from historical-era records (Schomburgk, 1848; now-extinct large tytonid owls apparently specialized Feilden, 1890; Harlow, 1925; Allen, 1942; Merrill, to feed on oryzomyine rodents (Steadman & Hilgart- 1958; Marsh, 1984, 1985; Lorvelec et al., 2007), and ner, 1999), the lack of native mammalian predators or all other taxa are known from Amerindian midden competitors throughout the region suggests that they deposits (Fig. 1; Table 4). The disappearance of the would have been extremely vulnerable to the intro- Lesser Antillean rice rats therefore constitutes part of duction of exotic mammals. The last known records the wider-scale series of land mammal extinctions for M. desmarestii, M. luciae, and O. victus all date across the West Indies that were driven by direct or from around the 1880s or 1890s (Table 4), and indirect human impacts during the historical era and although several authors have suggested that M. des- recent prehistory. However, the dynamics, drivers, marestii was wiped out by the eruption of Mount and timing of these extinctions remain poorly under- Pelée in 1902 (e.g. Allen, 1942; Balouet & Alibert, stood (Pregill & Olson, 1981; Morgan & Woods, 1986; 1990; Flannery & Schouten, 2001), it is more likely MacPhee et al., 1989; Turvey et al., 2007; MacPhee, that all of these extinction events on islands with long 2009; Turvey, 2009). volcanic histories were instead primarily driven by Rice rats were heavily exploited by prehistoric the introduction of the Indian mongoose (Herpestes Amerindians, and constituted a significant component javanicus) to the islands between 1889 and 1900 of their dietary intake (Wing, 2001a, b; Newsom & (Horst et al., 2001). However, it should be noted that Wing, 2004). Prehistoric human impacts on the M. desmarestii was already considered to be rare by region’s terrestrial environments escalated over time, some contemporary observers as early as 1820 (Lor- with substantial population growth as evidenced by velec et al., 2007). Most or all of the other Lesser an increase in the number and size of archaeological Antillean rice rat extinctions may have been caused sites occurring between AD 500 and AD 1000 (Keegan, by the accidental introduction of black rats across the 2000; Wilson, 1997, 2001; Newsom & Wing, 2004), insular Caribbean shortly after European arrival. and anthropogenic hunting pressures on rice rats and Although further data are required to clarify the other species may also have increased during this ecological mechanism(s) by which black rats and period. Some Amerindian sites show decreases in rice other exotic Rattus species act as extinction drivers, rat abundance over time (e.g. Mill Creek, Antigua; they have been implicated in the disappearance of Wing et al., 1968) or reductions in rice rat body size small mammals (including other insular rice rats) over time (e.g. Anse des Pères, St. Martin; Nokkert, and many other taxa on island systems across the 1999), supporting the hypothesis of prehistoric over- world through competition, predation, disease trans- exploitation and early preferential removal of larger mission, and habitat modification (Harris et al., 2006; individuals (cf. Lomolino et al., 2001). However, other Towns et al., 2006; Harris & Macdonald, 2007; Turvey archaeological sites show variable fluctuations in rice et al., 2007; Harris, 2008; Drake & Hunt, 2008; Wyatt rat abundance, which may reflect cultural or tapho- et al., 2008). nomic biases rather than true ecological patterns, and Although the taxonomy and species status of the rice rats typically persist into the youngest strati- Lesser Antillean rice rats remains extremely con- graphical levels and even increase in abundance over fused, the Windward and Leeward Islands alone have time in several sites (Jones, 1985; Newsom & Wing, lost approximately 20 separate island populations of 2004). There is little evidence that Amerindian envi- rice rats during the historical period, many of which ronmental impacts led to the prehistoric extinction of may have represented distinct species. This dramatic any West Indian rice rats, and radiometric dates level of extinction is equivalent in magnitude to the available for archaeological horizons from different much more widely known historical-era loss of islands show that many taxa definitely survived until marsupials and rodents in Australia (MacPhee & close to the time of first European arrival in the Flemming, 1999; Johnson, 2006; Turvey, 2009), but region around 500 years ago (Table 4). comprises only a part of the much greater series of West Indian terrestrial ecosystems have experi- recent land mammal extinctions documented so far enced a range of human impacts during the historical from the West Indies. This highlights the vulnerabil- era, including extensive forest clearance and associ- ity of insular mammal species, not only ancient ‘relict’ ated habitat degradation, and the introduction of a lineages (Purvis et al., 2000; Isaac et al., 2007), but wide range of exotic species. Identifying the primary also more evolutionarily recent insular radiations, causative factors for most Lesser Antillean rice rat and emphasizes the need for further investigation

© 2010 The Linnean Society of London, Zoological Journal of the Linnean Society, 2010, 160, 748–772 768 .T TURVEY T. S. Table 4. Last-occurrence dates for extinct rice rats from the Windward and Leeward Islands, based on historical records or calibrated radiometric dates from archaeological or palaeontological horizons containing rice rat material

Archaeological or palaeontological record

00TeLnenSceyo London, of Society Linnean The 2010 © Calibrated TAL. ET Historical radiometric Species Island record Site date, 2s Reference

Megalomys audreyae Barbuda Darby Sink AD 1173–1385 MacPhee & Flemming (1999) Megalomys desmarestii Martinique c. 1897 Allen (1942) Megalomys luciae St. Lucia pre-1881 Allen (1942) Oligoryzomys victus St. Vincent 1892 Allen (1942) Pennatomys nivalis gen. et Nevis 1720? (1930s?) Sulphur Ghaut AD 900–1200 Newsom & Wing (2004); Wilson sp. nov. (2006) St. Eustatius Golden Rock 80 BC–AD 980 Schinkel (1992) St. Kitts 1631? Bloody Point AD 660–1115 J.E. Robb, pers. comm. Undescribed taxon Anguilla Shoal Bay East AD 940–1320 Crock (2000) Undescribed taxon (?Megalomys Antigua Indian Creek AD 900–1100 Rouse & Morse (1999) olgclJunlo h ina Society Linnean the of Journal Zoological audreyae or ‘Ekbletomys hypenemus’) Undescribed taxon Barbados 1848? Schomburgk (1848), Ray (1962), Feilden (1890), Marsh (1984, 1985) Undescribed taxon Carriacou Grand Bay AD 390–1280 LeFebvre (2007); S.M. Fitzpatrick, (Grenadines) pers. comm. Undescribed taxon (large morph) Grenada Pearls 37 BC–AD 533 Haviser (1997) Undescribed taxon (small morph) Grenada Pearls 37 BC–AD 533 Haviser (1997) Undescribed taxon Guadeloupe Morel AD 21–881 Haviser (1997) Undescribed taxon La Desirade Petite Rivière AD 600–1400 de Waal (1996) Undescribed taxon Marie Galante Taliseronde AD 350–665 Haviser (1997) Undescribed taxon (large morph) Montserrat Trants 774 BC–AD 622 Petersen (1996) Undescribed taxon (small morph) Montserrat Trants 774 BC–AD 622 Petersen (1996) Undescribed taxon Saba Kelbey’s Ridge II AD 1290–1400 Hoogland (1996)

2010, , Undescribed taxon St. Martin Hope Estate AD 255–650 Newsom & Wing (2004)

Calibrated dates were calculated from conventional 14C ages (years BP) for Megalomys audreyae and for undescribed taxa from Grenada, Guadeloupe, and Marie 160 Galante using OxCal 4.0 (Bronk Ramsey, 1995, 2001). The calibrated date for M. audreyae is based on a direct radiocarbon date from subfossil rice rat material;

748–772 , all other radiometric last-occurrence dates are based on calibrated dates of stratigraphically associated material from archaeological sites. EXTINCT WEST INDIAN RICE RATS 769 and description of the region’s poorly known extinct tia (Muridae: Sigmodontinae) from a volcanic island off mammal fauna. Brazil’s continental shelf. American Museum Novitates 3256: 1–59. Crock JG. 2000. Interisland interaction and the development ACKNOWLEDGEMENTS of chiefdoms in the Eastern Caribbean. Unpublished PhD Detailed investigation of historical, zooarchaeological thesis, University of Pittsburgh. and palaeontological rice rat museum specimens was Crosby A. 2003. The prehistoric communities: the Hichman’s made possible by Irv Quitmyer, Sylvia Scudder, and landscape. In: Morris E, Leech R, Crosby A, Machling T, Elizabeth Wing (FLMNH-EAP), and by Andy Currant Williams B, Heathcote J, eds. Nevis Heritage Project interim and Louise Tomsett (NHM). Initial identiﬁcation of report 2002. Southampton: University of Southampton, 8–30. rice rat material from St. Eustatius was carried out Díaz-Franco S, Jiménez Vázquez O. 2008. Geocapromys by H.W. van der Klift. Further information and assis- brownii (Rodentia: Capromyidae: Capromyinae) en Cuba. tance with Caribbean archaeology and historical data Solenodon 7: 41–47. was provided by Peter Bellamy, John Crock, Andrew Drake DR, Hunt TL. 2008. Invasive rodents on islands: Crosby, John Guilbert, Jen Heathcote, Jim Johnson, integrating historical and contemporary ecology. Biological Quetta Kaye, Michelle LeFebvre, John Robb, and Invasions 11: 1483–1487. doi: 10.1007/s10530-008-9392-1 David Watters. Funding was provided for STT by a Evans C. 1968. The lack of archeology on Dominica. In: RP Leverhulme Early Career Fellowship and a NERC Bullen, ed. Proceedings of the Second International Congress Postdoctoral Fellowship; and for MW by an Alaska for the study of Pre-Columbian cultures in the Lesser Anti- EPSCoR Fellowship and a NSF (AToL) postdoctoral lles. Bridgetown, Barbados: Barbados Museum, 93–102. fellowship. Farris JS, Albert VA, Kallersjo M, Lipscomb D, Kluge AG. 1996. Parsimony jackkniﬁng outperforms neighbor- REFERENCES joining. Cladistics 12: 99–124. Feilden HW. 1890. Notes on the terrestrial mammals of Alcover JA, Campillo X, Macias M, Sans A. 1998. Mammal Barbados. Zoologist Series 3 14: 52–55. species of the world: additional data on insular mammals. Fischer JB. 1829. Synopsis mammalium. Stuttgart: JG American Museum Novitates 3248: 1–29. Cottae. Allen JA. 1912. Mammals from western Colombia. Bulletin of Flannery T, Schouten P. 2001. A gap in nature: discovering the American Museum of Natural History 31: 71–95. the world’s extinct animals. London: William Heinemann. Allen GM. 1942. Extinct and vanishing mammals of the Forsyth Major CI. 1901. The musk-rat of Santa Lucia (Anti- Western Hemisphere with the marine species of all the lles). Annals and Magazine of Natural History Series 7 7: oceans. Washington, DC: American Committee for Interna- 204–206. tional Wild Life Protection, Special Publication 11. Friant M. 1941. Notes anatomiques sur les Megalomys des Balouet J-C, Alibert E. 1990. Extinct species of the world. Antilles. Bulletin de la Société zoologique de France 66: New York: Baron’s Educational Series Inc. 122–129. Branch CW. 1907. Aboriginal antiquities in Saint Kitts and Friant M, Hoffstetter R, Chabanaud P. 1940. Appendice Nevis. American Anthropologist 9: 315–333. sur la faune précolombienne de l’Anse-Belleville. Journal de Bremer K. 1988. The limits of amino-acid sequence data in la Société des Américanistes 32: 259–274. angiosperm phylogenetic reconstruction. Evolution 42: 795– Gervais P. 1854. Histoire naturelle des mammifères, avec 803. l’indication de leurs moeurs, et de leurs rapports avec les Bremer K. 1994. Branch support and tree stability. Cladis- arts, le commerce et l’agriculture. Primates, cheiroptères, tics 10: 295–304. insectivores et rongeurs. Paris: Curmer. Bronk Ramsey C. 1995. Radiocarbon calibration and analy- Gullick MI. 1980. Changing Carib cookery. Proceedings of the sis of stratigraphy: the OxCal program. Radiocarbon 37: Eighth International Congress for the Study of Pre- 425–430. Columbian Cultures of the Lesser Antilles, 481–487. Bronk Ramsey C. 2001. Development of the radiocarbon Harlow VT, ed. 1925. Colonising expeditions to the West Indies calibration program OxCal. Radiocarbon 43: 355–363. and Guiana, 1632–1667. London: The Hakluyt Society. Burness GP, Diamond J, Flannery T. 2001. Dinosaurs, Harris DB. 2008. Review of negative effects of introduced dragons and dwarfs: the evolution of maximal body size. rodents on small mammals on islands. Biological Invasions Proceedings of the National Academy of Sciences of the 11: 1611–1630. doi: 10.1007/s10530-008-9393-0 United States of America 98: 14518–14523. Harris DB, Gregory SD, Macdonald DW. 2006. Space Carleton MD, Musser GG. 1989. Systematic studies of ory- invaders? A search for patterns underlying the coexistence zomyine rodents (Muridae, Sigmodontinae): a synopsis of of alien black rats and Galápagos rice rats. Oecologia 149: Microryzomys. Bulletin of the American Museum of Natural 276–288. History 191: 1–83. Harris DB, Macdonald DW. 2007. Interference competition Carleton MD, Olson SL. 1999. Amerigo Vespucci and the rat between introduced black rats and endemic Galápagos rice of Fernando de Noronha: a new genus and species of Roden- rats. Ecology 88: 2330–2344.

Haviser J. 1997. Settlement strategies in the Early Ceramic van der Klift HM. 1992. Faunal remains of the Golden Rock Age. In: Wilson SM, ed. The indigenous people of the site. In: Versteeg AH, Schinkel K, eds. The archaeology of Caribbean. Gainesville: University of Florida Press, 59– St. Eustatius: the Golden Rock site. Tilburg: ELKA Druk- 69. kerijen BV. Publication of the St. Eustatius Historical Foun- Hedges SB. 2006. Paleogeography of the Antilles and origin dation 2; Publication of the Foundation for Scientiﬁc of West Indian terrestrial vertebrates. Annals of the Mis- Research in the Caribbean Region 131, 74–84. souri Botanical Garden 93: 231–244. Kozuch L, Wing ES. 2006. Animal remains from archaeo- Hershkovitz P. 1944. Systematic review of the Neotropical logical sites on Nevis. In: Wilson SM, ed. The prehistory of water rats of the genus Nectomys (Cricetinae). Miscella- Nevis, a small island in the Lesser Antilles. New Haven, CT: neous Publications of the Museum of Zoology, University of Yale University Publications in Anthropology 87, Peabody Michigan 58: 1–101. Museum of Natural History, 145–186. Hershkovitz P. 1969. The evolution of mammals on southern Lambeck K, Chappell J. 2001. Sea level change through the continents. VI. The Recent mammals of the Neotropical last glacial cycle. Science 292: 679–686. region: a zoogeographic and ecological review. The Quarterly LeFebvre MJ. 2007. Zooarchaeological analysis of prehistoric Review of Biology 44: 1–70. vertebrate exploitation at the Grand Bay Site, Carriacou, Hershkovitz P. 1970. Supplementary notes on Neotropical West Indies. Coral Reefs 26: 931–944. Oryzomys dimidiatus and Oryzomys hammondi (Criceti- Lenglet G, Coppois G. 1979. Description du crane et de nae). Journal of Mammalogy 51: 789–794. quelques ossements d’un genre nouveau éteint de Cricetidae Higman BW. 2008. Jamaican food: history, biology, culture. (Mammalia-Rodentia) géant des Galapagos: Megaoryzomys University of the West Indies Press, Kingston. (gen. nov.). Bulletin de l’Académie Royale de Belgique Hoffman CA Jr. 1973. Archaeological investigations on St. (Classe des Sciences), 5th Series 65: 632–648. Kitts, W.I. Caribbean Journal of Science 13: 237–245. Lippold LK. 1991. Animal resource utilization by Saladoid Hoogland MLP. 1996. In search of the native population of people at Pearls. In: Ayubi EN, Haviser JB, eds. Proceedings Pre-Columbian Saba (400–1450 AD). Part Two: settlements of the Thirteenth Congress of the International Association in their natural and social environment. Unpublished PhD for Caribbean Archaeology. Curaçao: Anthropological Insti- thesis, University of Leiden, The Netherlands. tute of the Netherlands Antilles, 264–268. Hooijer DA. 1959. Fossil rodents from Curaçao and Bonaire. Lomolino MV, Channell R, Perault DR, Smith GA. 2001. Studies on the Fauna of Curaçao and other Caribbean Downsizing nature: anthropogenic dwarﬁng of species and Islands 9: 1–27. ecosystems. In: McKinney M, Lockwood J, eds. Biotic Hooijer DA. 1966. Fossil mammals of the Netherlands Anti- homogenization: the loss of diversity through invasion lles. Archives Néerlandaises de Zoologie 16: 531–532. and extinction. New York: Kluwer Academic/Plenum, 223– Hooijer DA. 1967. Pleistocene vertebrates of the Netherlands 243. Antilles. In: Martin PS, Wright HE Jr, eds. Pleistocene Lorvelec O, Pascal M, Delloue X, Chapuis J-L. 2007. Les extinctions: the search for a cause. New Haven: Yale Uni- mammifères terrestres non volants des Antilles Françaises versity Press, 399–406. et l’introduction récente d’un écureuil. Revue d’Écologie Hopwood AT. 1926. A fossil rice-rat from the Pleistocene of (Terre et Vie) 62: 295–314. Barbuda. Annals and Magazine of Natural History, Series 9 McCain CM, Timm RM, Weksler M. 2007. Redescription of 17: 328–330. the enigmatic long-tailed rat Sigmodontomys aphrastus Horst GR, Hoagland DB, Kilpatrick CW. 2001. The mon- (Cricetidae: Sigmodontinae) with comments on taxonomy goose in the West Indies: the biogeography and population and natural history. Proceedings of the Biological Society of biology of an introduced species. In: Woods CA, Sergile FE, Washington 120: 117–136. eds. Biogeography of the West Indies: patterns and perspec- McFarlane DA, Debrot AO. 2001. A new species of extinct tives, 2nd edn. Boca Raton, FL: CRC Press, 409–424. oryzomyine rodent from the Quaternary of Curaçao, Neth- Isaac NJB, Turvey ST, Collen B, Waterman C, Baillie erlands Antilles. Caribbean Journal of Science 37: 182–184. JEM. 2007. Mammals on the EDGE: conservation priorities McFarlane DA, Lundberg J. 2002. A Middle Pleistocene based on threat and phylogeny. PLoS ONE 2: e296. age and biogeography for the extinct rodent Megalomys IUCN. 2008. 2008 IUCN Red List of threatened species. Avail- curazensis from Curaçao, Netherlands Antilles. Caribbean able at http://www.iucnredlist.org Journal of Science 38: 278–281. Johnson C. 2006. Australia’s mammal extinctions: a 50000 McFarlane DA, MacPhee RDE, Ford DC. 1998. Body year history. Cambridge: Cambridge University Press. size variability and a Sangamonian extinction model for Jones AR. 1985. Dietary change and human population at Amblyrhiza, a West Indian megafaunal rodent. Quaternary Indian Creek, Antigua. American Antiquity 50: 518–536. Research 50: 80–89. Keegan WF. 2000. West Indian archaeology. 3. Ceramic Age. MacPhee RDE. 2009. Insulae infortunatae: establishing a Journal of Archaeological Research 8: 135–167. chronology for Late Quaternary mammal extinctions in the Kishino H, Hasegawa M. 1989. Evaluation of the maximum West Indies. In: Haynes G, ed. American megafaunal extinc- likelihood estimate of the evolutionary tree topologies from tions at the end of the Pleistocene. Dordrecht: Springer, DNA sequence data, and the branching order in Homi- 169–193. noidea. Journal of Molecular Evolution 29: 170–179. MacPhee RDE, Flemming C. 1999. Requiem æternam: the

last five hundred years of mammalian species extinctions. vertebrates in relation to Pleistocene climatic cycles. Annual In: MacPhee RDE, ed. Extinctions in near time: causes, Review of Ecology and Systematics 12: 75–98. contexts, and consequences. New York: Kluwer Academic/ Pregill GK, Steadman DW, Watters DR. 1994. Late Qua- Plenum, 333–371. ternary vertebrate faunas of the Lesser Antilles: historical MacPhee RDE, Ford DC, McFarlane DA. 1989. Pre- components of Caribbean biogeography. Bulletin of the Car- Wisconsinan mammals from Jamaica and models of Late negie Museum of Natural History 30: 1–51. Quaternary extinction in the Greater Antilles. Quaternary Purvis A, Agapow PM, Gittleman JL, Mace GM. 2000. Research 31: 94–106. Nonrandom extinction and the loss of evolutionary history. Marsh RE. 1984. Evidence does not support the existence of Science 288: 328–330. a Bajan mouse. Journal of the Barbados Museum and Ray CE. 1962. Oryzomyine rodents of the Antillean sub- Historical Society 37: 159–160. region. Unpublished PhD thesis, Harvard University. Marsh RE. 1985. More about the Bajan mouse. Journal of the Reig OA. 1987. An assessment of the systematics and evolu- Barbados Museum and Historical Society 37: 310. tion of the Akodontini, with the description of new fossil Merrill GC. 1958. The historical record of man as an ecologi- species of Akodon (Cricetidae, Sigmodontinae). Fieldiana: cal dominant in the Lesser Antilles. Canadian Geographer Zoology (New Series) 39: 347–399. 3: 17–22. Rouse I, Morse BF. 1999. Excavations at the Indian Creek Miller GS Jr. 1929. Mammals eaten by Indians, owls, and Site, Antigua, West Indies. New Haven, CT: Yale University Spaniards in the coast region of the Dominican Republic. Publications in Anthropology 82, Peabody Museum of Smithsonian Miscellaneous Collections 82: 1–16. Natural History. Miller GS Jr. 1930. Three small collections of mammals from Santori RT, Vieira MV, Rocha-Barbosa O, Magnan-Neto Hispaniola. Smithsonian Miscellaneous Collections 82: JA, Gobbi N. 2008. Water absorption of the fur and swim- 1–10. ming behavior of semiaquatic and terrestrial oryzomine Morgan GS, Woods CA. 1986. Extinction and the zoogeo- rodents. Journal of Mammalogy 89: 1152–1161. graphy of West Indian land mammals. Biological Journal of Schinkel K. 1992. The Golden Rock features. In: Versteeg the Linnean Society 28: 167–203. AH, Schinkel K, eds. The archaeology of St. Eustatius: the Musser GG, Carleton MD. 1993. Family Muridae. In: Golden Rock site. Tilburg: ELKA Drukkerijen BV. Publica- Wilson DE, Reeder DM, eds. Mammal species of the world: tion of the St. Eustatius Historical Foundation 2; Publica- a taxonomic and geographic reference, 2nd edn. Washington, tion of the Foundation for Scientific Research in the D.C: Smithsonian Institution Press, 501–755. Caribbean Region 131, 143–212. Musser GG, Carleton MD. 2005. Superfamily Muroidea. In: Schomburgk RH. 1848. The history of Barbados. London: Wilson DE, Reeder DM, eds. Mammal species of the world: Longman, Brown, Green & Longmans. a taxonomic and geographic reference, 3rd edn. Baltimore: Sorenson MD, Franzosa EA. 2007. Treerot, version 3. Johns Hopkins University Press, 894–1531. Boston, MA: Boston University. Newsom LA, Wing ES. 2004. On land and sea. Native Steadman DW, Hilgartner WB. 1999. A new species of American uses of biological resources in the West Indies. extinct barn owl (Aves: Tyto) from Barbuda, Lesser Antilles. Tuscaloosa & London: University of Alabama Press. Smithsonian Contributions to Paleobiology 89: 75–83. Nokkert M. 1999. Faunal exploitation. In: Hofman CL, Steadman DW, Pregill GK, Olson SL. 1984a. Fossil verte- Hoogland MLP, eds. Archaeological investigations on St. brates from Antigua, Lesser Antilles: evidence for late Martin (Lesser Antilles). The sites of Norman Estate, Anse Holocene human-caused extinctions in the West Indies. Pro- de Pères, and Hope Estate, with a contribution to the ‘La ceedings of the National Academy of Sciences of the United Hueca Problem’. Leiden: ASLU 4, 51–60. States of America 81: 4448–4451. Nokkert M. 2002. Preliminary report on faunal remains from Steadman DW, Ray CE. 1982. The relationships of Megao- 2002 excavations at Hichman’s (GE-5), Nevis. Unpublished ryzomys curioi, an extinct cricetine rodent (Muroidea: report, University of Southampton. Muridae) from the Galápagos Islands, Ecuador. Smithso- Olson SL, Maíz López EJ. 2008. New evidence of Ara nian Contributions to Paleobiology 51: 1–23. autochthones from an archeological site in Puerto Rico: a Steadman DW, Watters DR, Reitz EJ, Pregill GK. 1984b. valid species of West Indian macaw of unknown geographi- Vertebrates from archaeological sites on Montserrat, West cal origin (Aves: Psittacidae). Caribbean Journal of Science Indies. Annals of Carnegie Museum 53: 1–29. 44: 215–222. Stein BR. 1988. Morphology and allometry in several genera Patton JL, Hafner MS. 1983. Biosystematics of the native of semiaquatic rodents (Ondatra, Nectomys, and Oryzomys). rodents of the Galapagos Archipelago, Ecuador. In: Bowman Journal of Mammalogy 69: 500–511. RI, Benson M, Leviton AE, eds. Patterns of evolution in Swofford DL. 2001. Paup*. Phylogenetic analysis using par- Galapagos organisms. San Francisco: American Association simony (*and other methods), v. 4.0b10. Sunderland, MA: for the Advancement of Science, Pacific Division, 539–568. Sinauer. Petersen JB. 1996. Archaeology of Trants, Montserrat. Part Tate GHH. 1932a. The taxonomic history of the Neotropical 3. Chronological and settlement data. Annals of Carnegie cricetid genera Holochilus, Nectomys, Scapteromys, Mega- Museum 65: 323–361. lomys, Tylomys and Ototylomys. American Museum Novi- Pregill GK, Olson SL. 1981. Zoogeography of West Indian tates 562: 1–19.

Tate GHH. 1932b. The taxonomic history of the South Bulletin of the American Museum of Natural History 296: and Central American cricetid rodents of the genus 1–149. Oryzomys. – Part 2: subgenera Oligoryzomys, Thallomy- Weksler M, Percequillo AR, Voss RS. 2006. Ten new scus, and Melanomys. American Museum Novitates 580: genera of oryzomyine rodents (Cricetidae: Sigmodontinae). 1–17. American Museum Novitates 3537: 1–29. Thomas O. 1898. On indigenous Muridae in the West Indies; Wiens JJ. 1995. Polymorphic characters in phylogenetic sys- with the description of a new Mexican Oryzomys. Annals tematics. Systematic Biology 44: 482–500. and Magazine of Natural History 7: 176–180. Wilkinson M. 1995. Arbitrary resolutions, missing entries, Towns D, Atkinson I, Daugherty C. 2006. Have the and the problem of zero-length branches in parsimony harmful effects of introduced rats on islands been exagger- analysis. Systematic Biology 44: 108–111. ated? Biological Invasions 8: 863–891. Wilson SM, ed. 1997. The indigenous people of the Caribbean. Trejo-Torres JC, Ackerman JD. 2001. Biogeography of the Gainesville: University Press of Florida. Antilles based on a parsimony analysis of orchid distribu- Wilson SM. 2001. The prehistory and early history of the tions. Journal of Biogeography 28: 775–794. Caribbean. In: Woods CA, Sergile FE, eds. Biogeography of Turvey ST, ed. 2009. Holocene extinctions. Oxford: Oxford the West Indies: patterns and perspectives, 2nd edn. Boca University Press. Raton, FL: CRC Press, 519–527. Turvey ST, Grady FV, Rye P. 2006. A new genus and Wilson SM. 2006. The prehistory of Nevis, a small island in species of ‘giant hutia’ (Tainotherium valei) from the Qua- the Lesser Antilles. New Haven, CT: Yale University Publi- ternary of Puerto Rico: an extinct arboreal quadruped? cations in Anthropology 87, Peabody Museum of Natural Journal of Zoology 270: 585–594. History. Turvey ST, Oliver JR, Narganes Storde Y, Rye P. 2007. Wing ES. 1973. Notes on the faunal remains excavated from Late Holocene extinction of Puerto Rican native land St. Kitts, West Indies. Caribbean Journal of Science 13: mammals. Biology Letters 3: 193–196. 253–255. Vázquez-Miranda H, Navarro-Sigüenza AG, Morrone Wing ES. 2001a. The sustainability of resources used by JJ. 2007. Biogeographical patterns of the avifaunas of the Native Americans on four Caribbean islands. International Caribbean Basin Islands: a parsimony perspective. Cladis- Journal of Osteoarchaeology 11: 112–126. tics 23: 180–200. Wing ES. 2001b. Native American use of animals in the Voss RS. 1988. Systematics and ecology of ichthyomyine Caribbean. In: Woods CA, Sergile FE, eds. Biogeography of rodents (Muroidea): patterns of morphological evolution in a the West Indies: patterns and perspectives, 2nd edn. Boca small adaptive radiation. Bulletin of the American Museum Raton, FL: CRC Press, 481–518. of Natural History 188: 259–493. Wing ES, Hoffman CA Jr, Ray CE. 1968. Vertebrate Voss RS, Weksler M. 2009. On the taxonomic status of remains from Indian sites on Antigua, West Indies. Carib- Oryzomys curasoae McFarlane and Debrot, 2001, with bean Journal of Science 8: 123–139. remarks on the phylogenetic relationships of O. gorgasi Wing ES, Scudder S. 1980. Use of animals by the prehistoric Hershkovitz, 1971. Caribbean Journal of Science 45: 1– inhabitants of St. Kitts, West Indies. In: Lewenstein SM, ed. 7. Proceedings of the Eighth International Congress for the de Waal MS. 1996. The Petite Riviere excavations, La Desir- study of Pre-Columbian cultures in the Lesser Antilles. ade, French West Indies: ﬁeldwork report and subsistence Tempe: Arizona State University Anthropology Research studies for a pre-Columbian site with Late Saladoid and Papers No. 22, 237–245. post-Saladoid components. Unpublished MA thesis, Univer- Woods CA. 1989. The biogeography of West Indian rodents. sity of Leiden, The Netherlands. In: Woods CA, ed. Biogeography of the West Indies: past, Weksler M. 2003. Phylogeny of Neotropical oryzomyine present, and future. Gainesville, FL: Sandhill Crane Press, rodents (Muridae: Sigmodontinae) based on the nuclear 741–798. IRBP exon. Molecular Phylogenetics and Evolution 29: 331– Wyatt KB, Campos PF, Gilbert MTP, Kolokotronis S-O, 349. Hynes WH, DeSalle R, Daszak P, MacPhee RDE, Weksler M. 2006. Phylogenetic relationships of the oryzomy- Greenwood AD. 2008. Historical mammal extinction on ine rodents (Muroidea: Sigmodontinae): separate and com- Christmas Island (Indian Ocean) correlates with introduced bined analyses of morphological and molecular data. infectious disease. PLoS ONE 3: e3602.

SUPPORTING INFORMATION Additional Supporting Information may be found in the online version of this article: Appendix S1. Nexus ﬁle of morphological and molecular matrices. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.