WEST INDIAN XENODONTINE COLUBRID SNAKES: THEIR PROBABLE ORIGIN, PHYLOGENY, AND ZOOGEOGRAPHY VINCENT J. MAGLIO CONTENTS Al)stract 1 Introduction 1 A^e versus Haliitus 9 The Species Assemblages 9 cautJierigerus species assemblage 10 melanotus species assemblage 27 (ludrcae species assemblage 32 fiiiicreus species assemblage 36 A Problematical Cenus (laltiis) 48 Conclusions 48 Acknowledgments 50 Literature Cited 50 Appendix „_ 52 ABSTRACT The relationships between the thirty-three spe- cies of xenodontine snakes in the West Indies are reviewed primarily on the basis ot osteological and hemipenial morphology. Foiu" species assem- blages are recognized, distinguislied by the shape of the frontal and prefrontal bones and b\' the structure of the hemipenis. Within the caniheri- gcnis species assemblage three genera are recog- nized-A/.so/j/u's, HypsirJujnchus, and Uromacer. It is suggested that this group entered the West Indies from South or Central America, deri\ed from a primitive form of Alsophis. The South American species Alsophis chamissonis appears to be a relict of that primitive stock. The mainland genera Philoditjas and Conophis appear to be later specialized descendants from that same early stock. The three Galapagos species formerly referred to the genus Dromicits {hiserialis, dorsalis, and sk'vini), are placed in the genus AlsopJiis and regarded as close to the primitive mainland forms. The relationships of the genus laltris remain uncertain, but descent from West Indian Alsojiliis is reasonable. The mekinotits species assemblage has not progressed into the West Indies beyond the northern Lesser Antilles, and has almost certainly been derived from the mainland Leimadophis- Liophis-Lygophis complex. The generic name Dioniiciis is applied to these West Indian forms with the name Lcimadopliis as a junior synonym. Two species, andreae and parvifrons from Cuba and Hispaniola, respectively, share a number of osteological characters with Alsophis, but are like Dromicus externally. The hemipenis is of the Alsophis type and unlike that of Dromicus. Be- cause of their peculiar combination of characters these two species cannot readily be accommodated in any existing genus. The name AntiUopliis nov. gen. is proposed for them. It is suggested that they may be closely related to the mainland form Lijgophis hoursieri while the type species of that genus, L. lincatiis, appears to be closer to Dromi- cus. Eight species formerly placed in the genera Arrhytoii, Dromicus, and Darlingtonia, are con- sidered to form the funcrcus species assemblage. Except for the retention of Darlingtonia for hactiana, the species of this group are referred to the genus Arrhyton. A close relationsliip to the mainland genus Rhadinaca is postulated, and it is suggested that the two genera may have been derived from a connnon ancestor. The osteological similarities between these two groups are dis- cussed in terms of general seniiburrowing adapta- tions and are compared with other semi])urr()wiiig to burrowing New World colubrid snakes. It is concluded that these similarities represent a plnlo- genetic relationship rather than morpliological convergence. Four oversea colonizations from the mainland and numerous inter-island dispersals are recjuired to explain the recent West Indian fauna and its present distribution. INTRODUCTION Th(' \Vest Indies today contain an en- demic snake fauna of modest size. In the absence of an adequate fossil record, any Bull. Mus. Comp. Zool., 141(1): 1-54, December, 1970 1 BuUctiii Miisciiiu of Conipcnalive Zoology, Vol. 141, No. 1 Cuba I— SCALE 100 200 ' ■ MILES '° B] Great Inagua Anguilla \ Antigua S^Cro.x\ \Guadeloupe Puerto Rico %°"^\ \G '7 jAj ., .. , f^Dominica Nevisl V(. Montserrat/^j/ Marie Galante Martinique^'^St. OLucia Borbd^os Grenada, "^-0 Tobago Trinidad^ Fig. 1. Map of the West Indies in Mercator s projection. discussion of relationships within this group and of its history must ultimately be based on the inferred relationships of livinci; species. It is the purpose of the present study to examine the West Indian species of the subfamily Xenodontinae (sensii Dunn, 1928) of the family Colubridae, with reference to their origin, phylogeny, and zoogeography, so far as these can be de- duced from their anatomy and distribution. The only previous attempt to consider a large segment of this group was that by Dunn, 1932, but his work concerned only the Greater Antilles. Dimn relied heavily on the number of sensory pits present on each of the dorsal body scales and conse- quently recognized two basic generic groups in the West Indies; Ahophis was distinguished as having two pits per scale, and Dromicus only one pit. From these two groups Dunn derived all of the other endemic genera of the Greater Antilles. He also examined the dentition and hemipenis. concluding that these did not contradict his proposed relationships. However, ex- amination of Lesser Antillean and mainland species in the present stud}', as well as a re-evaluation of all West Indian xeno- dontines, does contradict these conclusions. Dimn weighted his characters in such a way that several well-defined groups of species remaincxl imrecognized. The xenodontine fauna of the \Vest Indies consists of the thirty-three species and their subspecies listed in Table 1. Tretanorhinus tjariahilis ssp. occurs on Guba, and appears to be a recent immi- grant from Gentral Anu^rica where se\'eral closely related species occur; it will not be considered further here. The remaining thirty-two species — except for ''Leima- doplm' mclanotus which occurs both on Trinidad and on th(> mainland (see Fig. 1 for map) — are endemic to the West Indies and form the basis of this work. All except I a It lis parislii ha\'e been examined. They West Indian Xenodontine Colubrid Snakes Maglio Table 1 Checklist and distribution of West Indian and some mainland^ and Galapagos^ xenodontine colubrid snakes. generic assignments recognized previous to this paper and recognized in the PRESENT STUDY ARE GIVEN FOR COMPARISON. SpECIES ARE ARRANGED IX ALPHABETICAL ORDER. Generic assignment Species Previous Present Distribution {almadensis) andreae andreae andrcac Diclopliyna andreae nehidatus andreae orientalis andreae ))eniu.stdae (an^iistilineatu-s) anomalus ater antillen.sis antillensis antillensis antiquae antillensis nianselli antillen.si.s sanetuntni antillensis sibonius ( bi.seriali.s) callilaennis cantherigerus cantherigerus canthcrigerus adspersus canth erigenis h rooksi cantherigerus caijmamts cantherigerus fiiscicauda cantherigerus pepei cantherigerus ruttiji cantherigerus sell wartzi catesbyi ( cluiniissiDiis) cursor dolichurum dorsulis dorsalis (dorsalis) exiguus cxiguus exiguus stahli exiguus subspadix ferox fremitus funereus haetiana haetiana liaetiana perfector Jul id e jidiae juliae co])eae jidiae mariae melanichnus melanotus ornatus (ixyrhynchus parishi parvifrons parvifrons Leimadophis Dromicus Drotiiicus Alsophis Alsopliis Alsophis Dromicus Dromicus Alsophis Uroniacer Dromicus Dromicus Arrhyton lultris Uromacer Dromictis Dromicus Ilypsirhynchus Uromacer Dromicus Darliugionia Dromicus Alsophis Leiniado])his Dromicus Uronuicer laltris Dromictis Dromicus Antillophis nov. gen. Alsophis Alsophis Alsophis Alsophis Alsophis Arrhyton Alsophis Uromacer Alsoiihis Dromicus Arrhyton laltris Uromacer Alsophis Arrhyton Ilyiisirhynchus Uromacer Arrhyton Darlingtonia Dromicus Also 1)1 1 is Dromicus Dromicus Uromacer laltris Antillophis nov. gen. Brazil Cuba Cuba Isle of Pines Cuba Cu])a Peru Hispaniola Jamaica Guadeloupe Antigua Montserrat Les Saintes Dominica Galapagos Jamaica Cuba Cuba Swan Island Grand Cayman Cayman Brae Cuba Little Cayman Cuba Hi.spaniola, Tortue Island, Vache Island, Gonave Island Chili, Argentina Martinique Cuba Hispaniola Gonave Island Galapagos American Virgin Islands Puerto Rico Puerto Rico Hispaniola Hispaniola, Beata Island Jamaica Hispaniola Hispaniola Dominica Guadeloupe Marie Galante Hispaniola Trinidad, Tobago, Soutli America St. Lucia Hispaniola Hispaniola Hispaniola ^ Non-West Indian species are enclosed in parentheses. Bulletin Miisciiiti of Coinparative Zoology, Vol. 141, No. 1 Table 1 ( Continued ) Generic assignment Species Previous Present Distribution parvifrans ciUeni parvifrou.s- lincolni IKirvifroii.s /i/^'cr l)(inifroiis jxiKiiii^cr parv if runs prat en us parvifron.s- rosanionde panifrons sii/giiis pdnifions tortugensis perfuscus po]\llepi.s ])()itoriccn.sis poitoiiccnsis- poiioriccnsis ancgadac puifoiircnsis a})luinl\is lioitoiiccii.si.s niclmtsi pcrtoriccn.si-s pn/nntii.s purtonccnsis richardi portoriccnms varicgatiis rijger.sntdi nifiroitiis sancticntcis (slevini) (tachymenoidcs) tacniatiun variahilis vHtdfitni fUtatinu vittdttnii Idiidoi vudii I' lid a I lid a allciiiiiiis vudii picticep.^ vudii rainetji vudii utawanae Lciiiiddopliis Droniicu.s Also)}his Ahophis Alsophis Also])hi.s Droiiiicus Drondcus Arrhijton Trctdnorhinus Arrhijton Alsophis Droviicus Arrhijton Alsophis Alsophis Also I )h is Alsophis Also I )h is Ahophis Arrhyton (not considered) Arrhijton Alsophis Gonave Island Hispanio]a Hispaniola Hispaniola Hispaniola Vachc Island Hispaniola Tortue Island Barbados Jamaica Puerto Rico Anegadae Vieques Buck Island Puerto Rico St. Thomas Mona Island Anguilla Bank Islands St. Kitts, Saba, St. Eustatius, Nevis St. Croix Galapagos Peru Cuba Cuba Cuba Cu1>a Gri'at Baliama Bank Islands Grand Bahama Island Bimini Island Crooked Island Great Inagua Island arc divided into four species assemblages on the basis of skull, heniipenial, and external characters, and will be treated within thes(> groups in the following pages. All described subspecies except "Dioiiicus' amireae pen/n.sfi/ae and "D." o. meloplujrrlia from Cuba, "Dromicus" jiiliae copeoe from Guadeloupe, and "D." exis until probable relationships have been assessed and assemblages of probable generic \'alue can be recognized. Changes in nomenclature are made only where West Indian Xenodontine Colubrid Snakes • Maglio 5 s soc Fig. 2. Skull of Alsophis conf/iengerus (MCZ 44874) showing general relationships of bones in xenodontine colubrid snakes. A, dorsal view; B, lateral view; C, ventral view. Left tooth-bearing elements removed. Abbreviations: bo, basioc- cipitcl; bs, basisphenoid; d, dentary; ec, ectopterygoid; eo, exoccipital; I, frontal; m, maxilla; n, nasal; p, parietal; pi, pala- tine; pm, premoxiila; po, postorbital; pr, prootic; pri, prefrontal; ps, parasphenoid; pt, pterygoid; q, quadrate; s, stapes; sa, surangular; sm, septomaxilla; soc, supraoccipital; st, supratemporal; tc, trabecular canal; v, vomer. Approx. X 7. 6 Bulletin Museum of Comparative Zoology, Vol. 141, No. 1 necessary for consistency and for the logical implementation of generic concepts as developed here. In a final section I use the postulated relationships to suggest a possible interpretation of the zoogeographic history of what I consider to be the valid groups in the West Indies. It is my feeling that snake genera are oversplit and not comparable to genera in other reptilian groups. This is a subjective judgment, however, and certainly not con- sistent with the philosophy and usage of the majority of herpetologists working on snakes. Since it is certainly desirable that taxonomic usage within the West Indies conform to that customary elsewhere, I have conserx'atively retained generic names (e.g., Darliniifonia) though I believe them to be of limited usefulness. In one case I have raised to generic rank a species group which, on current taxonomic usage, cannot be accommodated within any other existing genus. As Darlington (1938) and Simpson ( 1956 ) have discussed, the islands of the Caribbean do not appear to haxe been connected with the mainland during the later Cenozoic and overseas migration best explains the available faunal evidence. This hypothesis seems best also to explain the present data. Characters utilized The choice of specific skull characters was made only after more than 200 skulls of West Indian and mainland species had been examined to determine which char- acters were least variable within a species and to discover which ones could therefore be used to infer possible relationships with other species. The characters so chosen include^ the following: 1. The number of teeth which, in several of the species assemblages recognized here, show trends of reduction or increase from species to species (see Figs. 6, 20, and 29 below and the appendix). 2. The frontal bones (see Fig. 2 for labeled skull) also show considerable variation in shape within the West Indies, but prove to have distinctive proportions in certain assemblages, long and narrow on one group and nearly s(|uare in another. Figure 3 shows the distribution of length/ width ratios for the frontal bones in 29 West Indian species. The four groups of species labeled A, B, C, and D represent the species assemblages recognized here on the basis of all characters combined. Nevertheless, even with respect to frontal shape alone, it is clear that assemblage C may be distinguished from assemblages A and D in having a frontal pair that is roughly as wide as it is long. In assem- blage A, most of the species have a frontal pair that is one and one-half to two times as long as wide. 3. The shape of the prefrontal bone was found to be very reliable in separating assemblages in most cases, but (juanti- fication of this character was difficult. In Figure 4 the same 29 species as in Figure 3 are compared with respect to the length/ width index for the prefrontal bone. Al- though assemblages B and C overlap) completely, both exhibit a considerably longer and narrower prefrontal than in assemblage A (Fig. 5). Although there is a certain degree of variation with respect to shape of the prefrontal and other bones, they remain sufficiently distinctiv^e in each group to be taxonomically useful. Other characters such as the parasphenoid width and skull proportions are discussed under each species assemblage. Fig. 3. Length/width indices for the frontal bone pair of 29 species of West Indian colubrid snakes. A, conther/gerus species assemblage; B, me/anofus species assemblage; C, lunereus species assemblage; D, andreae species assemblage. Hori- zontal line, observed range; vertical line, mean; open rectangle, ± one standard deviation from the mean; solid rectangle, 95 per cent confidence interval of the mean. West Indian Xenodontine Colubrid Snakes • Maglio o ooo oooooooooooo aD0>O — cviro'tio <£>r^oo o>p — cm J L o o o o to ^ in to cvJ cvi cvj cvj cvi cj oj J \ I I \ \ L canfherigerus N=I2 vudi! N = 6 ater n=3 anomalus N=2 • melanichnus n=I -ig^„^ — portoricensis n=9 -1- sancticrucis N=2 -i- rufivenfris N=2 1 — rijgersmai N=2 1 00 I - antillensis N=6 -^ ferox N = 4 catesbyi n=7 ^i^^m^^- oxyrhynchus N=2 ' dor sails N=4 — I frenatus N=6 -i _L melanotus N=3 -L perfuscus n=2 -L- ornatus N = 2 -L cursor N=2 —1— yu/Zoe N = 4 i- funereus n = 2 . polylepis N=2 callilaemus N=2 — L hoetiana N=3 — L. exiguus N=4 vittatum N=7 — I/, landoi N = 3 T D 1 taeniatum n=2 I andreae N=2 parvifrons N=i3 O "1 — r "1 — r T ^ ^ 1 1 1 ^ \ \ \ \ OOOoOOOOOOOOOOOOOOO oooo— cvjro^iocot^oooo — cji*J'*;^ are based on dissections in situ of the une verted organ. Terminology is after Dowhng and Savage (1960). It must be emphasized that in this study most measured parameters show various degrees of overlap in range for various species. This does not, however, lessen their value in attempting to recognize phylogenetic relationships through an analysis of moiphological similarities. If an insular series of closely related species has been successively derived by island- hopping in a linear sequence, we might expect any two adjacent forms to show a greater similarity, barring extreme special- ization, than the two geographically terminal species of the series. The characters used here are those that combine relatively little intraspecific vari- ability and enough variation between spe- cies to be useful in the study of intrageneric relationships. Certain characters such as an apical awn on the hemipenis, the shape of the prefrontal bone, or certain skull pro- portions appear to be constant within groups of species that show a close relation- ship in totality of characters combined, and are therefore considered to be of maximum value in indicating true affinities. Others, such as the number of pits on each dorsal body scale, have sometimes proved incon- sistent \\'ith the majority of other traits and have therefore been considered un- reliable at the le\'el of major groups. ^ AGE VERSUS HABITUS When attempting to establish relation- ships between extant forms based entirely on morphological considerations, it must be reasonably certain that differences are not the result of allometric phenomena. Likewise, it must be reasonably clear that apparent similarities are not the result of habitus rather than a close phylogenetic relationship. Juvenile specimens, when available for comparison with the corresponding adult forms, show a number of consistent differ- ences in the structure of the skull which appear to be related solely to size. The major differences may be summarized as follows : Juvenile skull Skull relatively wide Quadrate thin and triangular Supratemporal relatively short Crests low and rounded Postorbital small, non-projecting Orbital foramen very large Pterygoids short, not projecting beyond foramen magnum Maxilla relatively lightly built Bones of brain case thin Adult skull Skull relatively narrow Quadrate with rodlike shaft Supratemporal relati\'ely long Crests high and sharp Postorbital large, projecting Orbital foramen small Pterygoids long, projecting far beyond foramen magnum Maxilla relatively massive Bones of brain case thick It is clear that these characters should not ordinarily be given high taxonomic weight unless the comparison is between two species of comparable adult size. Similarities resulting from habitus adap- tations are more difficult to establish be- cause the mode of life of these species is so poorly understood, and also because the adaptive significance of certain characters, such as scale pits, is at present unknown. Character convergence resulting from habitus similarity can, however, be inferred if a large number of characters are studied together. This point will be discussed in greater detail under the funereus species assemblage below. THE SPECIES ASSEMBLAGES Four species assemblages have been dis- tinguished among the 32 species of West Indian xenodontine snakes here under study. These are characterized on the basis of a number of traits as follows: 10 Build ill Museum of Cotu para five Zoology, Vol. 141, No. 1 cantherigerus assem])lage inelanotus assemblage undrcae assemblage funereus assemblage CANTHERIGERUS SPECIES ASSEMBLAGE Included West Indian species: anomalus Peters, Hispaniola; antillensis ( = leu- comdasY Schlegel, Antigua, Montserrat, Guadeloupe, Les Saintes, Dominica; ater GossE, Jamaica; cantherigerus- Bibhon, Cuba, Isle of Pines, Swan Island, Grand Cayman, Little Cayman, Cayman Brae; cateshiji Schlegel, Hispaniola, Tortue Is- land, Vache Island, Gonave; dor.salis Dunn, Gonave; ferox Gunther, Hispaniola; fren- atus GiJNTHER, Hispaniola; mehnichnus Cope, Hispaniola; oxyrliynchus Dumeril AND BiBRON, Hispaniola; portoricensis Rein- HARDT AND LuTKEN, PucrtO RicO, Moua Island, Virgin Islands; rijgersmm Cope, Anguilla Bank Islands; rufiventris Dumeril AND BiBRON, Saba, St. Eustatius, St. Kitts, Nevis; .sancticrucis Cope, St. Croix; viidii Cope, Great Bahama Bank Islands, Great Inagua. Osteology. The present group of species may be distinguished from other West Indian xenodontincs by a number of skull features, the most characteristic of which is the shape of the prefrontal bone (see Fig. 5). Here this element is wide antero- posteriorly with a broad and strongly con- vex anterior edge. The lower margin of this anterior edge cun^es medially aboxe the lacrimal foramen, so that the latter opens anteroventrally. ^As discussed by Schwartz (1966: 178), Brongersnia's (1937) analysis of Schlegel's co- types and his choice of the Guadaloupe-like speci- mens as the lectotype of Ahophis antillensis have reduced the name leucomclas to the junior synonymy of antillensis. - Senior synonym of angulifer; see Smith and Grant, 1958. Within the assemblage, the species cantherigerus (Cuba) has the lowest num- ber of teeth, with an average dental formula of about 12 -t- 2 maxillary, 10 palatine, 26 pterygoid, and 19 dentary teeth (see Fig. 6 and the Appendix for vari- ation). The skull (Fig. 7) is long and the cranium is moderately deep dorsoventrally. The frontals are widest anteriorly where they make contact with the prefrontals and are strongly emarginated above the orbits. A short, stout postorbital bone is articulated in a deep notch on the parietal bone in such a way that a prominent flange or lateral extension of the parietal intervenes between the postorbital and the frontal bones (see Fig. 2). A weak, but clearly visible groove marks the dorsal midsagittal line of the parietal bone. The parasphe- noid, forming the midventral surface of the skull, is narrow and has a deep trabec- ular canal on each lateral surface. This groove extends from the orbital foramen to the nasal capsule. A dorsal extension of the parasphenoid bone above the trabecular canals separatees the two orbits, forming a thin intcrorbital partition.' The supratemporal is strong and curved, and extends some distance beyond the occiput. The cjuadrate is long and straight. The species vudii on the islands of the Great Bahama Bank does not significantly differ ostcologically from cantherigerus. The dental formula is about 12 + 2 maxil- lary, 10 palatine, 24 pterygoid, and 21 dentary teeth for vudii vudii and is roughly comparable to that of cantherigerus. A peculiar feature of vudii is the melanic ^ Equals frontal crests of Underwood, 1967. West Indian Xenodontine Colubrid Snakes • Maglio 11 Fig. 5. Comparison of the three prefrontal bone types found in the four species assemblages of West Indian colubrid snakes as discussed in the text. A, conther/gerus assemblage type [Ahophis confher/gerus, MCZ 11200); B, me/onofus and andreae assemblage type (Antillophis parvi- frons, nov. gen., MCZ 77227); C, funereus assemblage type {Arrhyton polylepls, MCZ 81020). For each: left, lateral view; right, anterior view. Not to scale. Approx. X 10. tissue lining the cranial cavity of every specimen examined. This tissue imparts a bluish gray color to the skull. A similar condition occurs occasionally in cantheri- genis, and also in cateslnji, dorsalis, frenatus, and oxijrhijnchus where it is the usual condition. Although only two specimens including the type were available for study, vudii utowanoe from Great Inagua differs from all the subspecies of vudii on the Great Bahama Bank in several characters. The nasal bone is distinctive in shape, but this character appears to be somewhat more variable than most skull characters and is therefore of uncertain significance. In its dentition, however, vudii utowanae is quite distinct from the other subspecies. The dental formula is about 15 + 2 maxillary, 13 palatine, 30 pterygoid, and 23 dentaiy teeth, and is greater for every tooth-bearing element. When additional specimens of utowanae become available, a greater de- gree of overlap with the Bahama Bank forms may become evident. However, utowanae still will largely lie outside the neatly clustered range for the other sub- species of vudii. As in vudii vudii, the skull of utouanae appears bluish gray due to the melanic tissue lining the cranial cavity. On Jamaica the species ater has a dental formula higher than that of eantJierigerus. The skull is generally flatter (Fig. 8) and, as a result of this flattening, the nasal bones are closer to the septomaxilla, and the frontal bones touch the trabecular canals ventrally so that the interorbital partition is very small, consisting only of that portion of the parasphenoid bearing the trabecular canals. The frontal bones are relatively shorter and wider (Fig. 9) in contrast to the long, narrow frontals of the Cuban species. In all other skull characters the two forms are very similar. The septo- maxillae are expanded anteriorly and widen immediatelv behind the premaxilla (Fig. lOB). The name capistrata, introduced by Gosse (1851: 373) for a patterned form from Jamaica, was synonymized with ater by Boulenger (1894) without comment. Two specimens of this form from the British Museum were made available to me for comparative puiposes. Both are smaller than typical ater and differ from it in a number of osteological characters which in other species are related to 12 Bulletin Miisciiiii of Coiujniiafivc Zoolofnj, Vol. 141, No. 1 MAXILLARY TEETH — cvjrO'j-mtflr-oooo— c^ff' — — — fsj OJtMCVJ 1 I I I I I I \ I L canfherigerus N = I2 — — vudii N = 4 • vudii utowanae N=2 • ater N = 4 • anomalus N-2 • melanichnus N-l ^—^—. portoricensis N=IO ^— sancticrucis N=2 — rufiventris N=2 — rijgersmai N-2 antillensis N=6 — ferox N=3 catesbyi N = 6 oxyrhynchus N-7 — dorsalis N=2 frenatus N = 4 1 — I — \ — rT~i — r~\ — n i i r~ — cjro^intON-oooO — cMto CVJOJOJM PTERYGOID TEETH f^ooo^O — cjro^iou)h-00CT)O — cvj cJcJcvjc\Jc\JcMc\Jc\Jc\JOJroroio I I I I I I \ I I I I I I I I I vudii canfherigerus vudii utowanae — ater • anomalus • melanichnus portoricensis sancticrucis rufiventris rijgersmai antillensis ferox -— catesbyi oxyrhynchus • dorsalis — frenatus I I I I I I I I I I I I h-oo0— cvjrO'^-ioUJt^ajaiO— cj cvJOJCvjoJCvJCNJOJoJcocvjfOtOro TT PALATINE TEETH N CO CT) O — cviro^ioUJh- _l I \ I I I I I canfherigerus — vudii — vudii utowanae ater — anomalus • melanichnus portoricensis — sancticrucis rufiventris rijgersmai antillensis • ferox — catesbyi oxyrhynchus . dorsalis . frenatus 1 — I — rn — rn — i i i i — i — r~cocn2z^[2i:2^I^ DENTARY TEETH cDf^ooo^O— cvJro^incor-ooo^O — ojro^io cvjc\JCVJcJojcocvJCMoJcvjror<^rOrOrotO I I I \ \ l_J I 1 I I I 1_J \ \ \ \ 1 L_ canfherigerus — vudii — vudii utowanae ater — anomalus melanichnus portorice nsis — sancticrucis rufiventris — rijgersmai antillensis — ferox catesbyi oxyrhynchus •^— dorsalis • frenatus ~\ I I I I \ — \ — \ — \ — \ — \ — \ — \ — \ — \ — \ — \ — I — I — r lOf^oooiO — cviro'^u^tDf^coo o — cjrO'j-io oJOJcvJCMCJOJOJcdOJCO rorOfOtOfOro Fig. 6. Observed ranges of variation in numbers of teeth on each tooth-bearing element for the 15 West Indian species of the contherigerus species assemblage. ontogenetic changes. These inehide rela- capistroto represents a juvenile stage ofj tively narrower frontals, a broad rounded ater, and we may follow Boulenger in cranium, low crests and ridges, and thin rc\garding the two as synonymous, cranial bones. It thus seems likely that On Hispaniola there are eight species West Indian Xenodontine Colubrid Snakes • Maglio 13 5 mm Fig. 7. Skull structure of Alsophis cantherigerus, MCZ 56429. A, dorsal view; B, lateral view; C, ventral view. that may be placed in the present species 28 pteiygoid, and 24 dentary teeth. The assemblage. In the rare form melanichmis\ posterior processes of the vomer are later- the numbers of teeth are higher than in ally expanded into flat plates, oval in shape either cantherigerus or ater- the dental when viewed from below. This character formula is 18 + 2 maxillary, 16 palatine, is not seen in any other West Indian 14 Bulletin Museum of Conipamtive Zoology, Vol. 141, No. 1 oooooooooooo r^oooo — coroT^intDN-QO J III' I I I I \ L^ cantherigerus N = I2 vudii N = 6 ^ ater N = 3 — I — anomalus N = 2 . melanichnus N= I -i^^^a^ — porforicensis N=7 — ' — sancticrucis N = 2 — I — rufiventris N=2 — I— rijgersmoi N=2 -• — fez-ox N = 3 cafesbyi N = 7 oxyrhynchus N=6 -L- dorsalis N = 2 J frenatus N = 4 anfillensis N=6 Fig. 8. Frontal bone width/preorbital skull depth indices for the 15 West Indian species of the conther/gerus species assemblage. Horizontal line, observed range; vertical line, mean; open rectangle, it one standard deviation from the mean; solid rectangle, 95 per cent confidence interval of the mean. xenodontine. In comparison to canthcri- (Fig. 11). The skull has numerous high gerus, the postorbital bone is wider in crests and ridges for muscle attachment, proportion to its length, and the frontal is but these crests appear to develop with relatively slightly shorter. The supra- positive allometry in most large speci- temporal is short and stout. mens of xenodontines and are almost Another Hispaniolan species, anomalus, certainly related to the great size of this is closer in many ways to cantherigerus species. In its dentition, anomalus shows (Cuba) than it is to melanichnus, espe- only slight modifications from the con- cially in the structure of the vomer, post- dition found in cantherigerus (Cuba), orbital, and supratcmporal. The skull is and has a formula of about 12 +2 maxil- proportionately slightly wider and more lar\% 8 palatine, 20 pterygoid, and 19 dorsoventrally depressed, anterior to the dentary teeth. orbit, than in the Cuban form. The pre- Also found on Hispaniola is the well- maxilla is a solid, heavy structure, semi- defined species ferox. (I include speci- circular in ventral view, and quite unlike mens from the southwest peninsula of that of other members of the assemblage. Haiti which consistently lack a loreal Ventrally, the parasphenoid is wider be- scale.) This form is remarkably like ater neath the orbits than in cantherigerus and (Jamaica) in dorsoventral flattening of forms only a short interorbital partition the skull, in the short, wide frontal, and West Indian Xenodontine Colubrid Snakes • Maglio 15 o o oooo OOOOOOo ooooo — cMrO'^ ir)CO>-oocDO — c\Jro^tnaturc> alone. However, it is clear from other characters that the use of scale pits to define major groups may result in over- splitting of otherwise closely related as- semblages. With the recognition that scale pits by themselves are useful as taxonomic characters within this group only at the species or species-group level, the West Indian species of the cantherigerus as- semblage (excluding Ilypsirhynchtis and Uromacer) may be considered congeneric with chamissonis, angiistilineatus, and tachymenoides from South America, and with dorsalis, hiserialis, and slevini on the Galapagos archipelago. Smith and Grant (1958) have shown that Bibron's (1843) type of Dromicus was cursor from Martinique. This is a form unrelated to South American "Dromicus," as I will show below. WUh the name Dromicus thus unavailable, the present as- semblage of species is referred to Adsophis with the type antillensis ( = leucomelas) Fitzinger. (See Brongersma, already cited above, for the identity of the name antillen- sis.) The suggested phyletic relationships be- tween the genera and species of the can- therigerus assemblage are shown in Figure 18. Zoogeography. Two lines of evidence indicate a western origin for thc^ cantheri- gerus species assemblage into the West Indies. First, Alsophis cantherigerus from Cuba is the Antillean species most similar to A. chamissonis of the mainland; this similarity is most notable in osteological characters and especially in the dental formula, which in these two species is the lowest of the whole assemblage. The Gala- pagos forms, likewise, have relatively few teeth, as do the suggested mainland deriv- atives PJiilodryas and Conopliis. Within the West Indies a general trend toward increased number of teeth is evident, espe- cially in the specialized arboreal species of Uromacer, and in the portoricensis species group. From the geographic distribution of dental formulae, it would appear that a low number of teeth is primitixe for the mainland ancestor of this assemblage. In other characters also, the more easterly distributed species show a greater diver- gence from the mainland forms, thus tend- ing to support the \'iew that the group entered from the ^\'est. The second line of evidence indicating a western origin is that no member of this West Indian Xenodontine Colubrid Snakes • Maglio 25 CO .CO ^ ^ 05 "co "^ -Q CO O O ■"::: c: to 5q C o ^ -XT ^ C O -S CD to tu A Anguilla Bank e Great Bahama Bank c Cuba G Guadeloupe, Dominica GA Galapagos H Hispaniola Jamaica Nevis, St. Kitts, etc Puerto Rico St. Croix sA South America — oversea colonization Fig. 18. Suggested phyletic relationships between species of the cantherigerus species assemblage and related genera. Short horizontal lines indicate proposed oversea colonizations. Geographic distributions as indicated by lettered symbols. species asseml)lage occurs on the Lesser Antilles south of Dominica. This in itself is not significant since, as Gorman and Atkins ( 1969 ) have shown for Anolis, colonization does not necessarily proceed sequentially island by island along the chain of the Lesser Antilles. Nevertheless, taken with the first line of evidence, this assumes more importance. The following zoogeographical histoiy for the genus AIsoplus may be suggested. The assemblage appears to have been de- rived from an ancestral species probably not unlike Alsophis canflwrigenis in its osteological, hemipenial, and external mor- phology. From this widely distributed ancestral group in South America (and probably Central America as well), a single trans-Caribbean colonization could have resulted in the establishment of this group 26 Bulletin Museum of Comparative Zoology, Vol. 141, No. 1 A.vudii J A.cantlierigerus SCALE 100 200 I I I MILES A.rijgersmai ^portoriceQs^''"^""""' A.sancticrucisM A.ruflventri%^. 0. Fig. 19. Proposed routes of colonization of the West Indies by the cantherigerus species assemblage. The arrows are not intended to represent exact paths. The distribution of extant species and genera is as indicated. It is not certain whether this group entered from Central or South America. on Cuba. This was followed by a sub- sequent dispersion and radiation to Jamaica (ater), the Little and Great Bahama banks (vii(Ui), and Hispaniola (melanichnus) (Fig. 19). From Hispaniola an early stage of melanichnus gave rise to portoricensis on Puerto Rico. The four speeies of the Virgin Islands and Lesser Antilles, sancticrucis, lufiventris, rijp,er.smai, and antillensis, ap- pear to be part of a relatively reeent radi- ation of portoricensis, with differentiation on these geographically isolated islands. The remaining history of the group involves a series of inter-island colonizations to centrally placed Hispaniola from the more peripheral islands of the Greater Antilles. A second migration from Cuba by an early stage of cantheri^ienis seems best to ac- count for anomalus, \\'hich shows a closer affinity to the Cuban species than to melanichnus (Hispaniola). Alternatively, a back colonziation from Puerto Rico may have resulted in anomalus. The rather peculiar Hypsirhynchiis ferox is close to ater (Jamaica) with respect to its skull and hcnnipenis and possibly represents a Jamaica-Hispaniola migration, with sub- sequent specialization of Hypsirhynchtis resulting in its external differences. Another endemic genus on Hispaniola, Uromacer, may have been derived from an early form of H. ferox before the latter achieved its peculiar specializations. Because of its central location, Hispan- iola has received a greater number of species than any of the other islands, and it may have been competition between these species that led to the specializations of some of them. The giant form anomalus, the arboreal species of the genus Uromacer, and Hypsirhynchus may have differenti- ated as a means of dixiding up the habi- tat more efficiently. Competition among closely related and overlapping species West Indian Xenodontine Colubrid Snakes • Maglio 27 MAXILLARY TEETH t\JOJCJCJOJ(\Jf\JCJCJ I I I \ I I I L_ I I I — melanotus n=3 — perfuscus N=2 — or not us H-Z — cursor u-2 N=4 n — \ — \ — \ — I — \ — I — I — \ — \ — I — \ — \ — r OJCJCVJCVJCkJCJCVJCVJCVJ juliae PTERYGOID TEETH O — ojro^io uJNajOvO — (NPO'tio cvjcvjrocvjcvjoj cjcvjcNjcOfOrotororoto J \ \ L J LJ 1_L melanotus perfuscus 'ornatus — cursor juliae -| — I — \ — \ — I — I — \ — \ — r _ cvjojojcvjcvjcj cvJCMCMOjrOrorofOrorO n — I — \ — I — TT" O) O — CJ ro ^ PALATINE TEETH o — cvjio^io o — cvjro cvj rofororo J I L LJjll I I 4V — melanotus — perfuscus . ornatus — cursor 1 — r—\ — r I I im ih juliae _____ cvjrorOrOfO DENTARY TEETH (Dr-oocnO— cjfO^iouJh-oocno— cvi ro^ cvioj cJCvicvJCJCJcvJCNJCNJrorOfororo I I I I L_L I I I I _LJ L_L melanotus perfuscus .ornatus • cursor — juliae -1 — \ — [—] — I — I — m — I — n — I I I I \ n r ____c>jcvjrvjcvj(\jojcvjc\jcvjcvjrotorof0f0 Fig. 20. Observed ranges of variation in numbers of teetfi on each tooth-bearing element for the five West Indian species of the melanotus species assemblage. often results in the well-known phenom- enon of charaeter divergenee whereby the competmg forms become adapted ( first ecologically, then morphologically) to somewhat different aspects of the environ- ment, thus reducing competition. It is interesting to note that, even in the rela- tively less specialized Hispaniolan species anomalus and melonichnus, character di- vergence has progressed to a remarkable degree, so that these forms lie near the t\\ o extremes of variation for the entire assem- blage in many of their characters (see Figs. 6, 8, 9, and 11). Another waif dispersal from the main- land to the Galapagos almost certainly resulted in the differentiation of Imerialis, (lorsalis, and sJcvini on these islands. Later, rear-fanged specialization within the an- cestral mainland genus pbssibly led to Philochyas and Conophis which displaced their antecedent from much of its former range. The three closely related species, chamissonis, taclujmenoides, and ongusti- Uneofus, have remained as relatively un- modified relicts of the original mainland stock, except for reduction in the number of scale pits. MELANOTUS SPECIES ASSEMBLAGE Included West Indian species: cursor Lacepede, Martinique; julioe (including marlae) Cope, Guadeloupe, Marie Galante, Dominica; melanotus Shaw, South America, Trinidad, ? Grenada; ornatus Gaeman, St. Lucia; po-juscus Cope, Barbados. 28 Bulletin Museum of Comparative Zoology, Vol. 141, No. 1 Fig. 21. Skull length width incJices for the five West Indiarn species of the melanotus species assemblage. Widths are taken at the otic region of the skull. Osteologxj. This assemblage of species is distributed from the island of Guadeloupe south to Trinidad. It is distinguished from the Alsophis cantherigerus species assem- blage and other Antillean xenodontines primarily by the shape of the prefrontal bone and b\' the structure of the hemipenis (discussed below). The prefrontal is long dorso^'entrally and narrow anteroposteriorly with a sharply pointed anterior projection at about midlength (see Fig. 5). The species for which this group is named, melanotus, occurs on Trinidad and Tobago and has been questionably re- ported from Grenada. It also occurs in Venezuela and Colombia. Like all mem- bers of this group melanotus is about half the size of A. cantherigerus and contrasts with it in the following skull characters: the postorbital region is proportionately longer; the frontal bone is relatively shorter anteroposteriorly with very little emargin- ation above the orbits; the skull is con- siderably more flattened dorsoventrally as compared with its width; as a result of the latter character, the interorbital partition formed by the parasphenoid bone is shorter in its dorsoventral extension; long, thin lateral processes extend back from the premaxilla in contrast to the short, ^^ide based processes of A. cantherigerus. The dental formula for melanotus is about 15 + 2 maxillary, 10 palatine, 24 pterygoid, and 19 dentary teeth (Fig. 20 for variation). The skull of the endemic species per- fuscus on the island of Barbados is pro- portionately longer and narrower than that of melanotus (Fig. 21). The supratemporal and quadrate bones are more elongated and comparatively narrower. The orbit is small. The postorbital bone lies far for- ward on the parietal and is nearly in con- tact with the frontal. In this last character perfuscus is distinct from other members of the present group; in the latter, a wide expansion of the parietal bone separates the postorbital from the frontal. In general skull proportions and in its low dental formula, perfuscus is closer to melanotus and other mainland representatixes of this assemblage than it is to other Antillean species on adjacent islands. The species ornatus from St. Lucia is slightly larger than melanotus (Trinidad) and has a somewhat broader skull. The prefrontal bone is of the melanotus type, but is \\'ider than in that species. In most other skull characters the two forms are very similar except for the dental formula, which is higher in ornatus. Clearly related to the latter is the species cursor from Martinique (Fig. 22), which differs only in the smaller size of the nasal bone. On Guadeloupe, Dominica, and Marie Galante the species juliae has a higher dental formula than cursor, and the supratemporal, quadrate, and premaxilla are greatly re- duced in relative size. A similar reduction is seen in some related mainland species such as bimaculatus, but this probably represents parallelism rather than an inde- pendent derivation from one of these main- land forms. It is suiprising that no member of this group has been reported from St. Vincent. This island forms an important intermediate stepping stone between Grenada and St. Lucia. The mongoose is widespread on this island, and it is possible that a formerly present species, endemic or not, has be- come extinct. External morphology. Externally the melanotus species assemblage is a homo- West Indian Xenodontine Colubrid Snakes • Maglio 29 5 m m J I 1 — \ — I Fig. 22. Skull structure of Dromi'cus cursor, MCZ 6011a reversed). A, dorsal view; B, lateral vievv; C, ventral view. 30 Bulletin Museum of Comparative Zoology, Vol. 141, No. 1 Fig. 23. Hemipenis morphology in Dromi'cus cursor (MCZ 6011) showing the typical apical awn of the melanofus species assemblage; semidiagrammatic. The organ is un- everted and dissected in situ. Approx. X 5. geneous one and in many respects is similar to the cantherigerus assemblage. As in the latter group there are eight supralabials, but here only the fourth and fifth enter the orbit. The number of scale rows may be 17 or 19, and the ventral scales number appro-ximately between 150 and 200. In contrast to Alsophis, the number of sub- caudals is rarely over 100. The anal plate is divided. All of these snakes are moder- ate in size and are roughly between 600 and 1000 millimeters in total body length. A single apical pit is usually present on the dorsal body scales, but may be absent as in jiiliae mariae. Hemipenis. The hemipenis of cursor ( Fig. 23 ) is shorter proportionately than in Alsophis cantherigerus, and the sulcus spermaticus is less deeply divided. As in that species, several rows of stout spines extend along the sides of the organ but, in contrast to it, small spines are also present between the diverging branches of the sulcus. The organ is generally weakly bifurcated and the apical ornament is dis- tinct from all other West Indian xenodon- tines. Here papillae are never present; instead, a series of membranous folds radi- ate from the apex and terminate in a transverse fold of tissue which encircles the tip of each lobe. This fold forms a well-defined apical disk on each lobe when the organ is everted. The sulcus forks and proceeds onto the disk and to the tip of each lobe. The structure of the hemipenis in the other species in this assemblage is essen- tially like that of cursor. Origin and Zoogeography. The melanotus group offers no problem of origin. This well-defined and closely related assem- blage is moiphologically continuous with the widespread series of species currently referred to the genus Leirnadophis, common on the mainland.^ In both its osteology and hemipenial morphology the type species of Leimadophis — L. almadensis — is clearly al- lied to the present group, and there are no external characters which would pre- clude such a relationship. It seems probable that a northward colonization of the Lesser ^ Dromicus anuizouicus is known only from the type specimen, MCZ 2820, and may also be re- ferred to the present group. Its prefrontal bone is essentially of the melanotus type and the hemi- penis has a weak apical disk. In all its osteological characters this specimen may be distinguished from other members of the present group only in ha\ing those features, such as broad skull, tliin cranial bones, low, rounded crests, etc., which normally characterize juvenile specimens. Even the relatively weak disk on the hemipenis probably reflects the youthful condition of the specimen. It is thus quite possible that this form represents an immatmc specimen of a previously described South American species of the present assemblage. West Indian Xenodontine Colubrid Snakes • Maglio 31 SCALE 100 200 MILES D. perfuscus Fig. 24. Proposed routes of colonization of the West Indies by the melanotus species assemblage. Distribution of extant species is as indicated. Antilles by a mainland species similar to melanotus occurred in relatively recent times. From this species or its ancestor, ornatus (St. Lucia) was almost certainly derived, possibly via a now extinct inter- mediate species on St. Vincent. Successive overseas migration (Fig. 24) probably re- sulted in cursor on Martinique and jtilioe on Guadeloupe, Dominica, and Marie Galante. The Barbados species, perfuscus, appears to have been derived from one of the islands to the west, perhaps from St. Vincent or Grenada. Taxonomy. As mentioned above. Smith and Grant (1958) have sho\\'n that Bibron's (1843) type of Dromicus was cursor. With cursor and ahnadensis here considered as congeneric, Dromicus Bibron 1843 and Leimadophis Fitzinger 1843 become syno- nymic names for this assemblage. Although the actual dates of publication of these two names remain in question, December 31, 1843, is now to be taken as the official publication date of Fitzinger's Sijstema Reptilium (Smith and Grant, 1958), and Bibron's Dromicus thus becomes the senior synonym for the present assemblage with Dromicus cursor as the type species.^ The genus Dromicus is vei-v similar to the South American genus Liophis Wagler 1830. In all of the characters studied, Wagler's type of Liophis — L. cobella — is close to the present assemblage. The pre- frontal is like that of D. melanotus, the frontal is short with very little emargination above the orbits, and the interorbital par- tition is veiy small. The hemipenis of Liophis has a pair of well-developed apical disks as in D. melanotus, but differs in the presence of basal hooks (Roze, 1964). Body ^ After the present paper was in manuscript form, it was brouglit to my attention that Drs. Albert Sehwartz and Richard Thomas reached the conchision of "Dromicus" cf. cursor - Leimadophis synonymy independently of the present autlior. Their conclusion was reported in a letter to Dr. E. E. Williams. 32 Bulletin Museum of Comparative Zoology, Vol. 141, No. 1 .O Ci B Barbados D Dominico L Lesser Antilles M Martinique s St Lucia SA South America — oversea colonization Fig. 25. Suggested phyletic relationships between species of the me/onotus species assemblage and the genus Liophis. Short horizontal lines indicate proposed oversea coloniza- tions. size and scale counts are within the range of Dromicus (present concept). The question arises as to the vaUdity of the generic distinctions between ^'Leima- dophis" (i.e., Dromicus) and Liophis made by Roze (1964: 535). As we have seen, the presence or absence of scale pits may not be as important a distinction as for- merly belie\'ed. Thus, the only major difference between these two genera is in the maxillaiy dentition in which "Leim- adophis" has a diastema with much en- larged posterior teeth, whereas Liophis lacks a diastema and the posterior teeth are little enlarged ( Roze, 1964 ) . I suspect that these two groups will be considered as congeneric when better known, but on present evidence I here treat them as valid genera. The inferred phyletic relationships be- tween Liophis and the West Indian species of Dromicus are presented in Figure 25. ANDREAE SPECIES ASSEMBLAGE Included West Indian species: amlreae Reinhaedt and LxJTKEN, Cuba; parvifrons Cope, Hispaniola. OsteolofS,y. Only two species of this assemblage are extant in the West Indies, andreae on Cuba and parvifrons on Hispan- iola and nearby islands. They are generally considerably smaller in size than Alsophis and are about the size of Dromicus mela- nofus. In cranial osteology amlreae and parvifrons show features characteristic of both Alsophis and Dromicus (Fig. 26). The frontal bones are very long and narrow with a deep emargination above the orbits as in Alsophis, but unlike the pro- portionately shorter and \\ ider frontals of Dromicus (see Fig. 3). On the contrary, in the structure of the prefrontal bone, the present assemblage is close to Dromicus with a long and narrow prefrontal bearing a shaiply rounded anterior process. This is (juite distinct from the relati\'ely wider prefrontal of Alsophis in \vhich the an- terior surface is broadly rounded and the bone is proportionately wider (see Fig. 5). In general skull proportions andreae and parvifrons are closer to Alsophis than to Dromicus. The parasphenoid bone is very narrow as in A. cantheriiierus and other westerly distributed species of Alsophis, as well as mainland forms, but unlike the rather broad shape of that bone in the A. portoricensis species group or in Dromi- cus. The parasphenoid partition between the orbits extends high abo\e the trabecular canals, as in most species of Alsophis except West Indian Xenodontine Colubrid Snakes • Maglio 33 5mm I I I I L Fig. 76. Skull structure of Antillophis parvilrons nov. gen., MCZ 77228. A, dorsal view; B, lateral view; C, ventral view. 34 Bulletin Museum of Comparaiwe Zoology, Vol. 141, No. 1 Fig. 27. Hemipenis morphology in Aniillophis parvitrons nov. gen., MCZ 60064; semidiagrammatic. The organ is uneverted and dissected in situ. Approx. X 5. portoricensis and related species. The j)re- niaxillary bones in andreae and pawifrons lack the long lateral process as in Dromicus. The dental formnlae in these two species is comparable to that of both Alsopliis and Dromicus; that is, about 16 + 2 niaxillaiy, 12 palatine, 26 pterygoid, and 21 dentary teeth in parvifrons, and 21 + 2 maxillary, 16 palatine, 35 pterygoid, and 26 dentaiy teeth in andreae. Nonosteological characters. In external characters andreae and pawifrons are similar to Dromicus. The number of ven- tral scales is generally lower for these species than in Alsopliis and about the same as in Dromicus. The subcaudal scale number in parvifrons is higher than that of andreae and of species of Dromicus, while it is within the normal range for species of Aho})]}is. In the present assemblage, each dorsal body scale bears a single sensory pit in contrast to the two pits in all West Indian species of Alsophis, and in this respect is like most species of Dromicus. It is in the structure of the hemipenis, however, that andreae and parvifrons de- part radically from Dromicus. Here the organ has a deeply dix'ided sulcus sperm- aticus, bordered by a thick fold of spinose tissue. Several rows of stout spines parallel the sulcus, each branch of which terminates on a diskless apex bearing long filiform calyces (Fig. 27). There are no basal hooks, neither are there spines between the branches of the sulcus spermaticus as in Dromicus. This structure is basically like that of Alsopliis and is quite unlike that of Dromicus in \\'hich strong apical disks are present. Taxonomy and Origin. As noted above, the two species of the andreae group ex- liibit osteological features characteristic of l)oth Alsophis and Dromicus, though closer to the former. In most external characters they seem closer to Dromicus, but in the structure of the hemipenis they differ markedly from that group, being extremely close to Alsophis. Taken as a whole, this group cannot easily be referred to either genus. With regard to both skull and hemi- penial characters, these two species re- semble certain members of the South American genus Lygophis. The hemipenis in members of the latter genus, as currently recognized, is very heterogeneous. The organ may possess well-differentiat(>d apical disks as in L. lineatus (Fig. 2SA), the type species, and L. flavifrenatus, or it may be essentially of the Alsophis type, as in L. hoursieri (Fig. 2SB), in which the apical calyces are more spinulate along their margins than in the Alsophis type. Of these three species, Lygophis hoursieri (Ecuador West Indian Xenodontine Colubrid Snakes • Magjio 35 Fig. 28. Hemipenis morphology in the South American genus Lygophis; semldiagrammatic. The organ Is uneverted and dissected in situ. A, Lygophis lineatus, MCZ 80994; B, Lygophis boursieri, MCZ 36948. Approx. X 5. and Colombia) is very close to andreae and parvifrons in both skull and hemipenial characters, as well as in external scale pat- tern. From this incomplete study of Lygophis., it seems possible that we may be dealing with a compound genus of distantly re- lated forms. The evidence seems to sug- gest that L. boursieri might be placed in a separate genus with andreae and parvi- frons as a specialized radiation, perhaps derived from mainland Alsophis; and that the type of Lygophis, L. lineatus, may need to be synonymized with Dromicus from which it differs primarily in th(> lack of scale pits and the presence of basal hooks on the hemipenis (Roze, 1964). However, we must await more detailed taxonomic work to clarify these issues.^ Nevertheless, it seems clear that generic distinctions be- tween Dromicus (present concept) and ^ In a reappraisal of South American snakes related to LijgopJiis boursieri, Myers ( 1969 ) established the boursieri species group including Rhadinaea antioquien,sis, R. tiistriata, and L. boursieri. The group is placed in Lijgophis "... simply because boursieri already resides there." But the artificiality of the resulting genus is clearly pointed out, and the necessity of generic reassignment is affirmed. 36 Bulletin Museum of Comparative Zoology, Vol. 141, No. 1 Lygopliis (see Roze, 1964) rest on slim evidence. Since on present generic concepts it is not possible to accommodate the species of the ondreae group in any existing genus, it seems best to propose a new generic name. Though close similarities exist be- tween andreae and parvifrons and at least part of what is now called Lijr form a square plate above the orbits, unlike the condition in other West Indian groups. Ventrally the parasphenoid is proportion- ately wider along its entire length, especially posteriorly. The skull is pro- portionately flatter, and thus the inter- orbital partition formed by the dorsal extension of the parasphenoid does not extend above the trabecular grooves. Rather, the frontal bone on each side covers the entire lateral aspect of the para- sphenoid. The supratemporal and cjuad- rate are reduced in size, and the latter is flat and triangular in shape. On Jamaica there are three species of this group, funereus, polylcpis, and calli- lacmus. The specific distinction between funercus and polylcpis has recently been demonstrated by Buden (1966), but osteo- logically they are very similar. They have the highest dental formulae of the group (Fig. 29) with about 19 + 2 maxillary, 11 palatine, 19 pter\'goid, and 24 dentaiy teeth in funercus, and 17 + 2 maxillaiy, 11 palatine, 24 pterygoid, and 27 dentary teeth in polylepis. The parasphc>noid bone beneath the orbits is broad throughout its West Indian Xenodontine Colubrid Snakes • Maglio 37 MAXILLARY TEETH ,^oocnO-cM;o^|ncDf-ooo^o-cvj I I I I I 1 I I I I I • funereus N=2 • polylepis N - 2 — colli laemus N-Z — haetiana N-3 exiguus N=4 vittafum N=7 — V. landoi N-3 • dolichurum N = l taenia turn N = 2 1 I r 1 — \ — \ — I — I — r (-ooo^2z£d:25:!e^^?^g^^ PTERYGOID TEETH -r.mOOJ'a-CDOOOCVJ'd-lOCD yJCDii OJCJOJCMCJ I I I I I I I I I I I I • funereus — polylepis . callilaemus haetiana —^^—— exiguus — vittatum • / landoi • dolichurum — taenia turn T1 — \ — I — r I I I I U)Q0r:!:^:z.__c\jCJOJcM(\J PALATINE TEETH iou)t^condently in many groups vmder the in- fluence of similar habitus selection. But the exact way in which parallel characters are achieved, even under identical selection West Indian Xenodontine Colubrid Snakes • Maglio 45 pressures, depends upon raw materials in the form of existing moqohological struc- tures, and upon genetic variability. The more distantly related any two forms are, the more likely it is that they will achieve functionally similar adaptations in a di- vergent way. An examination of semiburrowing adap- tations in xenodontine snakes of the New World shows similar osteological trends common to all of them, but, as expected, they differ from each other in details. In most osteological characters studied, Rha- dinaea and the funereus group exhibit a similarity of form which suggests more than morphological parallelism with re- spect to semiburrowing adaptations. The osteological modifications which generally appear to be associated with semiburrowing adaptations are: small body size; reduction in relative orbit size; short- ening of the parietal region of the skull; enlargement and consolidation of the pre- orbital bones to form firm contacts with each other; broadening of the parasphenoid bone associated with the reduced orbits; relative broadening of the otic region so that it becomes the widest part of the skull (probably associated with general stream- lining ) ; reduction of the supratcmporal and quadrate bones; and, a relatively low num- ber of teeth. With respect to all of these characters, as well as others not obviously correlated with burrowing, the funereus group and Rhodinoea show a close correspondence, differing only in the structure of the pre- frontal bone. Tlie frontal bones form a nearly square plate above the orbits and contact the parietals in a broad, nearly straight suture. In contrast, the parietals of Geophis miititorqids extend lateral to the posterior half of the frontals, occupying a deep groove in the latter. A similar con- dition exists in ApostoJepis am])imii,ra, where the frontals are trapezoidal in shape, being wider anteriorly. In Atractus Jati- frons the frontals are oval in shape, with their long axis in the transverse direction. In the funereus group and Rhadinaea the parietal retains the shape normal for nonburrowing xenodontines. In Geophis mutitorquis a posterior process extends into the deeply divided interparietal, and in Atractus lotifrons and Rhinostoma guia- nense the parietals are so shortened that they are broader than long. The reduced quadrate in the funereus group and Rhadinaea is thin and tri- angular in shape. In Rhinostoma two forms appear to be relatively West Indian Xenodontine Colubrid Snakes • Maglio 47 C3 O Q: to to 03 CD c Cuba CA Central America H Hispaniola J Jamaica p Puerto Rico V Virgin Islands — oversea colonization Fig. 36. Suggested phylefic relationships between the species of the funereus species assemblage and the genus Rhad- inaea. Short horizontal lines indicate proposed oversea colonizations. recent products of speciation on that island. The most primitive member of the genus is A. funereus which forms the base of both the A. funereus-A. toeniatiim series and the A. funereus-A. exiguus series. It is the only Antillcan species (except for A. pohjiepis) with a nonbifurcated hemipenis and in this respect is closest to species of Rliadinaea. Within the A. funereus-A. taeniatum series it is the least specialized in terms of the reduction of skull bones and other semi- burrowing adaptations. From this early A. funereus stock two basic lines appear to have been established; A. coIUkiemus (Jamaica) began a trend toward reduction of the posterior skull bones and in the dentition, and a widening of the para- sphenoid bone. A colonization of Cuba from this early A. calliloemus stock led to the Cuban forms which, through repeated speciation, continued the trend, with A. taeniatum as the most highly specialized species of the group. The second line re- sulted from a migration to Hispaniola and tended toward a narrower parasphenoid, larger size, and in many ways a general convergence toward other W'cst Indian xenodontine groups, especially Dromicus. This may have been related to a general tendency away from semiburrowing adap- tations. In order to derive A. exiguus from this early Hispaniolan form we must postu- late a widespread distribution for this latter form before (or concurrent with) its differ- 48 Bulletin Museum of Comparative Zoology, Vol. 141, No. 1 SCALE 100 200 I I 1 MILES D. haetiana A . funereus ^ ■ exiguus ^^ A . poly le pis Fig. 37. Proposed routes of colonization of the West Indies by the funereus species assemblage. The arrows are not in- tended to represent exact paths. The distribution of extant species is as indicated. entiation into the morphologically some- what specialized and ecologically restricted species Darlingtonia haetiana. Today D. haetiana occurs only in the montane massifs of the southwestern and Barahona penin- sulas at altitudes ranging from 1000 to 5600 feet. One possible explanation for the peculiar distribution of this species is an ecological replacement of the former wide- spread species (possibly by pawifrons?) with D. haetiana remaining as a montane relict. This zoogeographic pattern is sum- marized in Figure 37. A PROBLEMATICAL GENUS Two species remain to be discussed: laltris (lor.salis and 7. parlshi from Hispan- iola. laltris dorsalis (I have not seen I. pari.shi) Ls a large species and is most similar to Alsophis in many skull char- acters, but distinct in many external and hemipenial characters. The skull is nar- rower, especially in the otic region. In its dentition this species is unique among West Indian xenodontines in having prominent grooves on the enlarged posterior maxillary teeth. The bilobed hemipenis is very long, ridged with numerous folds, and has an apical ornament of weakly developed flounces. Externally there are seven up- per labials as in Arrhijton generally, but the \'entral and subcaudal scale counts are similar to those of Alsophis. laltris dorsalis (and presumably I. parishi also) is not very close to any other Antillean species as far as can be determined from its present morphological specializations and therefore certainly should be retained in a distinct genus. In most characters laltris shows its greatest similarity to Alsophis, and it may have been derived from that genus on Hispaniola. CONCLUSIONS The use of skull and hemipenial char- acters, in addition to those of external West Indian Xenodontine Colubrid Snakes • Maglio 49 Table 2 Distinguishing charactebistics of the eight genera of xenodontine colubrid snakes in the West Indies as discussed in the text. type Hijpsirhijnchus cantherigeriis long and no disk type narrow Uromacer cantherigerus long and no disk type narrow large no divided 1 present lalt ris cantherigerus long and no disk type narrow large, 8 no divided present arboreal large 7 yes dixided present to be of significant suggesting relationships between morphology, appears aid in species of West Indian xenodontine eohi- brid snakes. They not only provide data for a proposed redefinition of generic con- cepts, but suggest certain phylogenetic relationships with mainland groups. Such relationships are of considerable interest, since they allow a tentative reconstruction of the possible origin and history of these snakes in the Antilles. The generic groups of xenodontine snakes here recognized in the West Indies and listed in Table 1 may be distinguished as in Table 2. The present xenodontine fauna of the West Indies was possibly wholly derived from Central and South American stocks through at least four oversea colonizations. Based on present evidence, a summaiy of the postulated historical events follows: 1) From the formerly widespread South American genus Alsophis, a waif coloni- zation established this group on Cuba. Sub- sequent radiation into a number of species and endemic genera led to its present distribution throughout the Greater An- tilles and the northern Lesser Antilles. A minimum of three separate inter-island migrations of this group is required to explain the peculiar faunal assemblage of Hispaniola. 2) The specialized genus laJtris possibly emerged from Alsophis on Hispan- iola. 3) Using Jamaica as a port of entry and center of dispersion, a single stock, possibly derived from the Central Ameri- can genus RhacUnaea, successfully spread through the Greater Antilles in two distinct but closely related lines and established the genera Arrliyton and Dorlingtonia. 4) Pos- sibly derived from part of what is now called Lijg,oph\s in Soutli America, the species andreae and parvifrons may have reached Cuba and Hispaniola by a direct oversea colonization. 5) A relatively recent invasion of the Lesser Antilles by a species of Dromicus ( = Leimadophis) almost cer- 50 Bulletin Museum of Comparative Zoology, Vol. 141, No. 1 tainh' (Mitered vhx Trinidad, but has not yet progrt^ssed be>'ond (iiiadcloupe. The chronological sequence of coloni- zations cannot definitely be established on present evidence. However, a sequence roughh' similar to that above is not un- reasonable. It, of course, cannot be as- sumed that the West Indies were devoid of a xenodontine ophi fauna before the series of colonizations that established the present fauna, but our knowledge of earlier xenodontine colonizers is nonexistent be- cause of the lack of a significant fossil record. From the patterns of dispersion discussed in this paper it would appear that numerous combinations of inter-island migrations have occurred. The main sequences have progressed from one island to the next adjacent island and in this sense were for the most part linear. The following series have been proposed: mainland-Cuba- Hispaniola-Puerto Rico-Lesser Antilles; mainland-Cuba-Jamaica-Hispaniola; main- land-Cuba-Bahamas; mainland-Jamaica- Cuba; mainland-Jamaica-Hispaniola-Puerto Rico; mainland-Trinidad-Lesser Antilles. Inter-island migration, especially to cen- trally located Hispaniola, seems to have been more frec^uent than mainland-island migrations. This was certainly the result of the greater cross-water distance between the mainland and any island than between the various islands themselves, as Simpson (1956) and Darhngton (1957) have sug- gested. The greatest diversity in species and genera occurs on Hispaniola; this is to be expected in view of its central position and consequently greater number of coloni- zations. Its large size, varied habitats, complex physiography and history have prox'ided an excellent opportunity for im- migrants to differentiate into noncompet- ing forms. The zoogeographical patterns here pro- posed are based on limited evidence and are in large measure speculative. It is hoped that they offer a workable contri- bution toward the continued study of this group. However, only when adequate in- formation about the comparative anatomy, karyotypes, ecology, physiology, and bio- chemistry of all Antillean snakes and their mainland relatives is available will we be able to draw firmer conclusions concerning the origin and zoogeography of xeno- dontine snakes in the West Indies. ACKNOWLEDGMENTS I wish to express my deep gratitude to Professor Ernest E. Williams for suggesting the problem which began this work, for innumerable discussions and suggestions, and for his constant encouragement. With- out his interest and assistance this study could never have been completed. For their reading of various versions of the manuscript and for many useful com- ments and criticisms, I extend my thanks to Drs. Richard Estes, George Gorman, Max Hecht, Edmond Malnate, Charles Myers, James Peters, Albert Schwartz, Richard Thomas, Paulo Vanzolini, and Ernest Wil- liams. The conclusions reached in this paper and any errors which remain are, of course, solely my responsibility. I am grateful to the following persons who kindly loaned specimens from their collections and ga\e permission to extract skulls: Dr. Richard Zweifel, Miss A. G. C. Grandison, Drs. Edmond Malnate, Albert Schwartz, and Ernest Williams. I wish thankfully to acknowledge Mr. Laszlo Meszoly \\'ho prepared the maps, graphs, and skull drawings, and Miss Catherine McGeary and Mrs. B. Gail Browne who typed various versions of the manuscript. The research was supported in part by National Science Foundation grants nos. GB-6944 and NSF B 019801X. LITERATURE CITED Barbour, T., and B. Shrevk. 1938. Concerning some Bahamian reptiles with some notes on' the fauna. Proc. Boston Soc. Nat. Hist., 40 (5): 347-366. BiBROK, G. 1843. Reptiles, in Ramon De La Sagra, Histoire Physique PoHtique et Natur- elle de I'ile de Cuba, pp. 1-239. West Indian Xenodontine Colubrid Snakes • Maglio 51 BouLENGER, G. A. 1894. Catalogue of the snakes in tlie British Museum (Natural History). London, vol. 2: 1-382. . 1896. Catalogue of the snakes in the British Nhiseum (Natural History). London, vol. 3: 1-727. Broxgersma, L. D. 1937. The types of Psam- moi)lii.s antillensis Schlegel, 1837. Heipeto- logical Note XIV, Zoologische Mededeelingen, 20: 1-10. BuDEN, D. W. 1966. An evaluation of Jamaican Dromicus with the description of a new species. Breviora, No. 238: 1-10. Cochran, D. M. 1941. The Herpetology of Hispaniola. Bull. Smithson. Inst., U. S. Nat. Mus., No. 177: 1-398. Cope, E. D. 1862. Synopsis of the species of Holcosis and Ameiva, with diagnoses of new W. Indian and S. American Colnbridae. Proc. Acad. Nat. Sci., Philadelphia, 1863: 60-82. Darlington, P. J., Jr. 1938. The origin of the fauna of the Greater Antilles, with discussion of dispersal of animals over water and through the air. Quart. Rev. Biol., 13: 274-300. . 1957. Zoogeography. New York: John Wiley and Sons, 675 pp. Dowling, G. H., and J. M. Savage. 1960. A guide to the snake hemipenis: a survey of basic structure and systematic characters. Zoologica, 45 (1): 17-27. Dunn, E. R. 1928. A tentative key and arrange- ment of the American genera of Colnbridae. Bull. Antivenin Inst. Amer., 2(1): 18-24. . 1932. The colubrid snakes of the Greater Antilles. Copeia, 1932 (2): 92-98. Gorman, G. C, and L. Atkins. 1969. The zoo- geography of Lesser Antillean Atioli.s lizards- an analysis based upon chromosomes and lactic dehydrogenases. Bull. Mus. Comp. Zool., 138 (3): 53-80. Gosse, p. H. 1851. Naturalists' sojourn in Jamaica. London: Longman, Brown, Green, and Long- mans, pp. 1-508. Horn, H. 1969. Polymorphism and evolution of the Hispaniolan snake genus Uromaccr (Colnbridae). Breviora, No. 324: 1-23 Lanix), R. v., and E. E. Williams. 1970. Notes on the herpetology of the U. S. Naval Base at Guantanamo Bay, Cuba. Studies of the fauna of Curacao and other Caribbean islands. No. 116: 159-201. Mertens, E. 1939. Herpetologische Ergebnisse eine Reise nach der Insel Hispaniola, Westin- dien. Abh. Senckenberg. Naturforsch. Ges., No. 449: 1-84. Myers, C. W. 1967. The pine woods snake, Rhadinaca flacilata (Cope). Bull. Florida State Mus., 11 (2): 47-97. . 1969. South American snakes related to Lygophis hoiiisieii: a reappraisal of Rhadinaca antioquicnsis, Rhadinaca tristriata, Coronclla wJiymperi, and Liophis ataliiiallpae. Amer. Mus. Novitates, 2385: 1-27. RozE, J. A. 1958. A new species of the genus Urothcca (Serpentes: Colnbridae) from Vene- zuela. Breviora, No. 88: 1-5. . 1964. The snakes of the Leimadophis- Urotheca-Liophis complex from Parque Na- cional Henri Pittier ( Rancho Grande ) , Vene- zuela, with a description of a new genus and species ( Reptilia, Colnbridae ) . Senckenbergi- ana Biol., 45 (3/5): 533-542. Schwartz, A. 1965. Snakes of the genus Alsophis in Puerto Rico and the Virgin Islands. Studies of the fauna of Curacao and other Caribbean islands. No. 90: 177-227. . 1966. Review of the colubrid snake genus Arrhijton with a description of a new sub- species from southern Oriente Province, Cuba. Proc. Biol. Soc. Washington, 78: 99-114. Simpson, G. G. 1956. Zoogeography of West Indian land mammals. Amer. Mus. No\'itates, 1759: 1-28. Smith, H. M., and C. Grant. 1958. The proper names for some Cuban snakes: an analysis of dates of publication of Ramon De La Sagra's Historia Natural de Cuba, and of Fitzinger's Systema Reptilium. Herpetologica, 14: 215- 222. Underwood, G. 1962. Reptiles of the eastern Caribbean. Carib. Affairs (n.s.), Dept. Extra- mural studies, Univ. West Indies, No. 1 : 1-92. . 1967. A contribution to the classification of snakes. British Mus. (Nat. Hist.), Pub. No. 653: 1-179. 52 BuUciin Museum of Comparative Zoology, Vol. 141, No. 1 Appendix Dental pormxjlae and variation for West Indian and some related mainland and Galapagos xenodontine colubrid snakes. variation includes ranges of subspecies.'^ West Indian Xenodontine Colubrid Snakes • Maglio 53 AprEXDix { Continued ) N2 maxillarv palatine pterygoid dentary laliris dorsalis- Lio))Jiis- aiioDuila cohella jacgcri mcrrcmi Lijguphis hoursieri flavifrenutus liucatits Philodryas acsiivus Jntrmeisteri olfersii Uromaccr catcshtji dorsalis fremitus oxyrJiyucIiu.s Rhadinaca Incvirostrls d ecu rat (I flavdata ■serperaster 16 + 2 21 20 1 Problematical subspecies which may be full species are listed separately. - Number of specimens examined.