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

AMERICAN MUSEUM Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2853, pp. 1-12, figs. 1-8 September 15, 1986

An Enigmatic New Snake from the Peruvian Andes, with Notes on the Xenodontini (Colubridae: Xenodontinae)

CHARLES W. MYERS'

ABSTRACT Liophis problematicus, new species, is based on in some populations but which are accepted as a a specimen from 1520 m elevation on the Ama- defining synapomorphy pending further study. zonian side of the Andes in the Department of Defensive neck flattening or hood display is Puno, extreme southeastern Peru. It is a very small widespread in the Xenodontini, having been re- snake (adult male 275 mm total length), with a corded for at least five of the six genera currently bilobed hemipenis having nude apical discs, a con- assigned, and seems to provide a behavioral syn- dition that defines a probably monophyletic group apomorphy that corroborates the validity of the (tribe Xenodontini) of at least six genera of Neo- group. North American snakes of the genus Het- tropical xenodontine colubrids. Beyond this, the erodon also flatten their necks and have been com- relationships ofthe new species are uncertain, but pared morphologically with some Xenodontini; a a pragmatic assignment to Liophis (s.l.) scarcely relationship between Xenodon and South Amer- violates the definition ofthis highly variable, pos- ican Hydrodynastes, another genus ofneck spread- ers, also has been suggested. Heterodon and Hy- sibly nonmonophyletic genus. Incidental notes are drodynastes have relatively primitive hemipenes provided on the hemipenis, coloration, and be- and cannot be included in the Xenodontini. The havior ofLiophis williamsi, a small snake ofVen- hypothesis that one genus or the other is a sister ezuelan cloud forest. group ofthe Xenodontini deserves to be tested by It is pointed out in discussion that hemipenial morphological criteria-although this seems to be variation among the Xenodontini is much greater strongly contraindicated by immunological dis- than has been indicated in the literature, and that tance data. Neck spreading therefore may have the tribe is based essentially on a single character- evolved several times within the subfamily Xe- the paired apical discs, which may have been lost nodontinae.

1 Chairman and Curator, Department of Herpetology, American Museum of Natural History.

INTRODUCTION The main purpose of this paper is to de- nude apical discs (Xenodontini), Liophis scribe a new xenodontine colubrid snake from problematicus can be distinguished by the the Andes of southern Peru, and secondarily combination of small size (<300 mm total to call attention to a behavioral trait that length), low number of ventrals and subcau- seems to offer additional support for the tribe dals (133 + 36 in one male), and color pat- Xenodontini. The new species is quite small tern-including pale labial markings, pale (fig. 1 is life size), and I know it only from a nuchal spots, and a striped body (figs. 1-3). single specimen in the Field Museum ofNat- DISTRIBUTION: Liophis problematicus is ural History. known only from its type locality at 1520 m At first examination, the color pattern and elevation on the Amazonian side of the An- small adult size of the new snake reminded des (drainage of Rio Madre de Dios)-about me ofa few species ofRhadinaea. Hemipeni- 150 km north of Lake Titicaca in the border al morphology, however, indicates that it is region between Peru and Bolivia. allied with some member of a group or tribe of Neotropical genera that includes Eryth- DESCRIPTION rolamprus, Liophis, Lystrophis, Umbrivaga, Waglerophis, and Xenodon. But the new The holotype (fig. 1) is an adult male as species does not readily fit in any of these determined by enlarged and convoluted vasa genera, which for the most part appear to deferentia, enlarged segments of the kidney represent well-defined monophyletic units of tubules, and hardening of the hemipenial ecologically diverse snakes. spines. The specimen is in fair condition ex- Liophis, however, is not defined by syn- cept for a slightly misshapen mouth border apomorphy and accommodates snakes that in the rostral region and front of the lower are quite divergent in color pattern and mor- jaw, owing to desiccation. phology. Therefore, inasmuch as this genus PROPORTIONS AND SCUTELLATION: A small, is of questionable monophyly and in order slender, short-tailed snake 275 mm total to avoid creating or resurrecting a new ge- length, 51 mm tail length (tail 18.5% oftotal). neric name at this time, a provisional assign- Body higher than wide (roughly 6.5 mm x ment to Liophis is warranted. The specific 5.5 mm), rounded ventrolaterally. Head name derives from the Greek problematikos slightly wider than body; greatest head width (problematical), an adjective here used in ref- (parietal region) 6.5 mm, head length 11.3 erence to uncertain kinship. mm from tip of snout to level ofend of man- dibles. Diameter ofeye greater than distance Liophis problematicus, new species Figures 1-6 HOLOTYPE: FMNH 64733, an adult male obtained by Hilda H. Heller in the period November 22-December 20, 1950, at 1520 m above sea level at San Juan, Tambopata [Rio], Sandia [Province of ], Department of Puno, Peru. I assume that "San Juan" is the same as San Juan del Oro, a settlement situated about 30 km ENE of the town of San- dia, on the west bank of the Rio Tambopa- ta-at 14°13'S, 69°10'W.2 DIAGNOSIS: Among those snakes characterized by a bilobed spiny hemipenis with 2From Mapa Fisico Politico del Peru (1:1,000,000). Ministerio de Guerra del Perui, Instituto Geografico Na- Fig. 1. Liophisproblematicus, new species. The cional, 1982 (2nd ed.). holotype (FMNH 64733), shown at natural size. 1986 MYERS: NEW SNAKE FROM ANDES 3

Fig. 3. Midbody color pattern ofLiophisprob- lematicus, drawn from holotype on generalized scale pattern. Fig. 2. Dorsolateral view of snout of Liophis problematicus, showing transverse keel or crease on rostrum of holotype, x 40. Uppermost arrow supraocular, preocular, loreal, nasal, and in- is at midpoint ofanterior edge ofrostral plate. The ternasal (left prefrontal in contact with both keel may be an artifact (see fn. 3). left and right internasals). Frontal hexagonal (with small anterior apex), 2.0 times longer from its anterior edge to anterior edge of na- than its greatest width, 1.5 times longer than ris, extending 1.5 times into length of snout. distance from its anterior edge to tip ofsnout. Dorsal scales smooth, lacking apical pits and Supraoculars posteriorly more than one-half anal ridges, in 19-17-17 rows; scale-row re- of frontal width, narrowed anteriorly. Pari- duction from 19 to 17 rows by fusion of lat- etals about 1.7 times longer than broad; in- eral rows 3 + 4, on both sides, at level of terparietal suture shorter than length ofsnout, 62nd ventral. Two preventrals (undivided about 60 percent length offrontal plate. Large gulars), 133 ventrals, divided anal plate, 36 nasal plate, in contact with first two supra- pairs of subcaudals. labials, grooved apparently above and below Rostral plate visible from above, about 2.5 naris; small loreal plate ovoid, separated from times wider than high, showing a blunt hor- eye by the single high preocular (no subpreoc- izontal keel (fig. 2) that may be an artifact.3 ular); two postoculars, the upper twice the Paired intemasals, each wider than long, size oflower. One large anterior temporal and about three-fourths as long as prefrontals. upper and lower posterior temporals, be- Paired prefrontals, each wider than long, each tween labials 5-7. Supralabials 7, second in contact with its mate and with frontal, touching loreal, third and fourth bordering orbit. Infralabials 8, first pair in contact behind mental, 1-4 touching an anterior genial, 3 It seems likely that the blunt rostral keel is an arti- 4-5 touching a posterior genial. Genials well ficial fold or crease, owing to the front ofthe snout being developed; anterior pair slightly shorter than somewhat dried and misshapen and the rostral plate posterior pair but with longer intergenial su- being partially separated from the premaxillary bone. ture. Inconspicuous tubercles (presumed sen- Rostral modification in very small snakes most often is on an adaptation for semifossorial habits, but usually there sory organs) sparsely present head plates. are correlated specializations, including modified head COLOR PATTERN: Body light brown, with shields, reduced eyes, and a pointed head that is small darker middorsal brown (gray) streak and lat- relative to the neck. Liophisproblematicus, however, has eral and ventrolateral blackish brown lines a normally proportioned head and normal head plates (fig. 3), as follows: (1) The middorsal streak expected of an aboveground forager. is medium brown (gray after loss of stratum 4 AMERICAN MUSEUM NOVITATES NO. 2853

Fig. 5. Left maxilla of Liophis problematicus (holotype), x 1O.

Fig. 4. Head of Liophis problematicus (holo- There is no dark stripe through the eye. type), approximately x 3.9. The supralabials bear three conspicuous white markings that are broadly edged in dark brown, as follows: There is a short horizontal corneum) and extends from the nape to the white line across labials 1-3; a small subocu- end of the tail; this dark streak is five scales lar white area on labial 4; and a bold post- wide on the neck, three scales wide on the ocular white stripe that originates on the low- greater length ofthe body, and narrows grad- er postocular and extends obliquely across ually on the tail. (2) The dark lateral line labials 5, 6, and lower corner of 7 to the edge extends from the rear of the head to the tip ofthe mouth, where the line matches up with ofthe tail, becoming darker (brown to black- a white line on the anterior part of the last ish brown), more nearly continuous (broken (8th) infralabial. The underside of the head at anterior end), and widening perceptibly is basically white, with brown blotches on the from anterior to posterior. The lateral line mental and on each of the infralabials and lies on only the lower edge of scale row 6 genials and on many of the gular scales. anteriorly, then drops (because of scale row DENTITION: There are 15 (right side) or 16 reduction) onto row 5 before midbody; it (left) recurved subequal maxillary teeth, which covers the lower half of row 5 at midbody, are followed by a distinct diastema and two nearly all ofrow 5 toward the end ofthe body, ungrooved fangs that are about I1/2 times larg- and continues conspicuously along the side er than the prediastemal teeth (fig. 5). The of the tail, where it narrows only toward the ultimate prediastemal socket on each side lies end. (3) Brown pigmentation extends from anterior to the front edge ofthe ectopterygoid the lower sides onto the ventral and subcau- process. The fangs have a rounded anterior dal plates before ending abruptly at a blackish surface, a flattened knifelike posterior edge, brown ventrolateral line that extends from and are somewhat laterally compressed at the the head to the end of the tail. The median tips; the ultimate fang is offset slightly laterad. three-fifths of the ventral surface is immac- The numerous palatine-pterygoid teeth are ulate white in preservative, and sharply set not readily counted in situ. There are at least off by the dark ventrolateral line on either 26 teeth on the left dentary. side. VERTEBRAE: As examined ventrally with the The head is brown (gray under stratum cor- internal organs displaced, the hypapophyses neum) above, irregularly blotched with dark- ofthe cervical vertebrae appear to have a long er brown, and set off posteriorly by several ventral edge (i.e., broad in profile, not spine- white markings (fig. 4) as follows: (1) A small like), and the m. transversohypapophyseus is white spot immediately behind the interpa- present. Posteriorly, the hypapophyses are re- rietal suture occupies less than half the area duced to a lower, thick hemal keel and the ofthe first dorsal scale. (2) A pair ofobliquely m. transversohypapophyseus is absent (ex- arranged white nape spots lie one scale be- amined at level of ventrals 115-116 in area hind the parietal plates and are separated by of kidney overlap). the width of the brown vertebral scale row; HEMIPENIS: The right retracted organ ex- each of the large nape spots occupies all or tends in situ to the end of subcaudal 7, being parts of 6-7 scales. (3) A rounded white spot, bifurcated from about the middle of subcau- occupying parts of three scales, lies on each dal 7 and with the two slips of the major side of the neck immediately posterior to the retractor muscle merging toward the base of last supralabial and narrowly separated from subcaudal 8 (combined division ofhemipenis the nape spot above. tip and muscle only the length of about one 1986 MYERS: NEW SNAKE FROM ANDES 5

hemipenis is everted.) The sulcus spermaticus lies on the lateral wall and bifurcates slightly below the middle of the hemipenis, with the branches extending to the tips ofthe lobes and maintaining well-defined walls even on the apical discs; the branches ofthe sulcus have a centrifugal orientation, lying on opposite sides of the distal lobes (i.e., in the uneverted organ in fig. 6, one branch lies on the dorsal wall of the dorsal lobe and the other branch is on the ventral wall of the ventral lobe-a configuration that can be determined by manipulation during dissection as well as by inspection of everted organs [discussion in Myers and Campbell, 1981, pp. 16-17]). A basal naked pocket is situated dorsally on the proximal one-fourth of the hemipenis (in fig. 6, the distal part of this pocket is concealed by the overhanging spines). Approximately the basal 15 percent of the organ is nude, lacking spines or spi- nules. The rest ofthe organ is densely spinose, with several dozen medium-size spines on the midsection giving way to numerous small spines along the branches of the sulcus spermaticus and up to the edges ofthe apical discs (in fig. 6, small spines along the edge of the visible disc are concealed under its over- hang). DISCUSSION Fig. 6. Hemipenis of Liophis problematicus HISTORICAL SUMMARY: Assuming that I (holotype). Uneverted right organ opened mid- have correctly interpreted the morphology of ventrally, x 10. The proximal free edge ofthe api- its uneverted hemipenis,4 the relationships of cal nude area overhangs and conceals small spines that would surround the edge ofa presumably flat- 4 Morphological misinterpretations ofboth uneverted tened apical disc if the hemipenis were everted; a and everted hemipenes are commonplace in the litera- second disc is concealed in the unopened dorsal ture. The following examples are relevant: Thompson lobe (see fig. 7 for an idea of the probable config- (1913, p. 79) mistook as apical discs the distal nude areas uration when everted). on the sulcate sides of the lobes of an uneverted organ in Oxyrhopus trigeminus (lobes distally calyculate on asulcate sides, e.g., in AMNH 104667), and erected the subcaudal plate). The major retractor muscle invalid genus Erythroxyrhopus largely on that basis. And originates at the level of subcaudal 27. The Wellman (1963, p. 289) incorrectly described the everted right hemipenis was opened along its mid- organs of Conophis as being moderately calyculate with ventral surface and then removed and pinned tiny apical discs on the short lobes. The spinose hemi- out for detailed study and illustration (fig. 6). penis of Conophis has neither calyces nor discs, being The small lobes constitute only about the conspicuously flounced distally (flounces with spinulate distal tenth of the retracted hemipenis. The to papillate edges) and having a centrolineal sulcus spermaticus. The tips of most Conophis organs I examined tip of the opened ventral lobe comprises naked are not quite everted (AMNH 66337, 123917, 126423), distensible tissue that is set off by a free but, in the one illustrated by Wellman (UMMZ 82650), proximal edge. (This nude area and the one the apex of each short bilobation has at the terminus of concealed in the unopened dorsal lobe pre- the sulcus a hard, nearly nude (slightly papillate) bump sumably form flattened apical discs when the or projection-the mistaken disc. 6 AMERICAN MUSEUM NOVITATES NO. 2853

Liophis problematicus are most likely to be I think the above genera may well form a sought among the xenodontine genera that monophyletic group that can usefully be giv- possess bilobed spiny hemipenes with nude en tribal status. But it should be pointed out apical discs. Cope (1894, pp. 840, 842; 1895, that the group is recognized mainly by hemi- pp. 200-201, 207), although understandably penial resemblance, and that some of the misinterpreting some of his pioneering dis- characters which account for this resem- sections, recognized two assemblages ("dis- blance are probably plesiomorphic for colu- ciferi") characterized by the apical discs. But brids generally (e.g., bilobed hemipenis, bi- he separated these groups widely, placing furcated sulcus spermaticus, spines present); them in the 1895 paper as subfamilies ofsep- a few features not present (calcyces, capita- arate families and superfamilies5 -a separa- tion) might be symplesiomorphic for most tion mainly based on whether the posterior Xenodontinae but secondarily absent in the maxillary fangs were grooved or not. Dunn Xenodontini, although synapomorphies of (1928, p. 21) tried to bring the generic no- loss are especially difficult to confirm. menclature up to date, and gathered Cope's There has been no adequate comparative disciferi together as a presumably natural study ofhemipenial variation among the gen- subgroup of "Ophiinae" (=Xenodontinae) era of this group, or even among the species characterized by a "disked, double" hemi- of a single genus. And yet the variation is penis.6 H. G. Dowling has advocated tribal considerable: The apical discs range from status for this and other assemblages, and small to large relative to the size ofthe hemi- nine modem generic names representing penis; the branches of the sulci spermatici Dunn's "disked, double" group were listed vary in orientation from centrolineal to cen- under the tribe Xenodontini in Dowling and trifugal; the base ofthe organ goes from squat Duellman ("1974-1978" [1978], p. 112a.3). to fairly long; and the lobes, although usually Since then, Dixon (1980) has relegated three very short, may be very slender and much of the names to the synonymy of Liophis longer than the base. A particular problem in (Dromicus, Leimadophis, Lygophis), thus re- recognizing the Xenodontini is that the de- ducing the group to the following six genera: fining tribal character-the apical discs-appears to have been lost in some populations Erythrolamprus of Xenodon, the type genus (Myers, unpub- Liophis lished data)! Lystrophis With the above reservations and pending Umbrivaga additional study, the hemipenial apical discs Waglerophis provide a likely synapomorphy on which the Xenodon hypothesis ofmonophyly ofthe Xenodontini is accepted for present purposes. There may also be a behavioral synapomorphy (fig. 8 5Aglyphodonta, Colubridae, Xenodontinae; and Gly- phodonta, Dipsadidae, Erythrolamprinae. and accompanying text). Cadle's immuno- 6 Dunn (loc. cit.) removed Thompson's Erythroxyrho- logical data from 13 species representing five pus from this assemblage (see fn. 4), but he mistakenly of the six genera were consistent with the added Philodryas and Siphlophis. As explained by Dow- hemipenial synapomorphy and he regarded ling (1969, p. 5), addition of Philodryas probably re- the group as a clade, mentioning that his data sulted from the nomenclatural confusion in Cope's work. were insufficient for addressing questions of Dunn's placement of Siphlophis, however, is puzzling intragroup relationships (Cadle, 1984a, p. 17). because he indicated (by the suffix "D") that it was based The above summary skips over various at- on his own examination ofthe hemipenis. It seems most tempts to relate one or more of the genera likely that Dunn would have looked at the relatively with North American Heterodon, a view that common and widespread type species ofSiphlophis (Ly- was summarized and further supported by codon audax = S. cervinus), which clearly has the hemipenial lobes apically calyculate (e.g., in AMNH 116342, Underwood (1967, pp. 100-103), who point- uneverted) rather than disced. Conceivably he might have ed out, however, that one such claim had seen a partially everted hemipenis, in which case the been based on "slender and partly faulty evi- distal calyces might not be visible and the lobes would dence" and that there are discrepancies among appear short. the anatomical accounts ofseveral writers. In 1986 MYERS: NEW SNAKE FROM ANDES 7 common with the times and with other au- number of subcaudals as is Liophis proble- thors whom he cited, Underwood appeared maticus, and one species is as small as prob- to attach no taxonomic importance to wheth- lematicus; however, the species of Umbri- er characters are primitive or derived, giving vaga have stripes confined posteriorly on body more weight, for example, to the symple- and tail, with spots or bands anteriorly on siomorphic condition of "divided hemi- the body, and pale supralabials. The preced- penis" than to the unusual apical discs (op. ing genera also differ from L. problematicus cit., pp. 102, 148). Neill (1964) added the in various kinds of maxillary specialization. South American Dugandia and Cyclagras- Some of the characters alluded to above monotypic genera now subsumed within Hy- (as well as characters not mentioned) are drodynastes (see Dowling and Gibson, doubtlessly synapomorphic, and I find no 1970)-as possible relatives of Xenodon, grounds for assigning the new snake to any based mainly on the shared presence of a ofthe five considered genera. This leaves Lio- bifurcate sulcus spermaticus and distensible phis (sensu Dixon), which Dixon (1980, p. hood (Neill's primary objective was disso- 26) recognized as the "most generalized ge- ciating one species from an erroneous place- nus of the group in its osteology, external ment in the Colubrinae). morphology, habitat, and food preferences." If the hemipenial apical discs truly define As now constituted, Liophis is extraordinar- the Xenodontini as a monophyletic group, ily diverse: The species range in habitus from then Heterodon and Hydrodynastes clearly slender to relatively robust; adult total length cannot be included because they have rela- varies from small (e.g., <500 mm in L. wil- tively primitive hemipenes. Nonetheless, an- liamsi) to large (e.g., >1 m in L. miliaris); atomical similarities, such as claimed for the tail is short to long (tail/total length = Heterodon and some of the Xenodontini, 0.135-0.310 fide Dixon); color patterns in- should be reevaluated in the context ofa syn- clude unicolor, speckled, banded, and several apomorphy scheme in order to test the pos- distinctive types of striping, including com- sibility that Heterodon (or Hydrodynastes) + bined patterns such as anterior blotches and Xenodontini form a larger monophyletic posterior stripes. The rather extreme diver- group. Ifanatomical evidence ofa sister-group sity raises a suspicion that Liophis may be relationship were found, then certain behav- polyphyletic.7 Most of the dentitional and ioral resemblances might acquire new mean- other osteological traits used by Dixon to dis- ing, as mentioned at the end of this paper tinguish Liophis from related genera seem to (under A Behavioral Synapomorphy). Ca- be plesiomorphic, and it is perhaps the one dle's immunological data, however, predict genus of the group that lacks obvious syn- that such a close relationship will not be apomorphies. found; Hydrodynastes seems much closer to Consequently there is nothing in Dixon's several other South American lineages than (1980, p. 22) generic description that logically to the Xenodontini (Cadle, 1984a, p. 12), prevents problematicus from being assigned whereas the enigmatic Heterodon "seems far to Liophis (excluding from consideration a removed from all South American xenodon- few unexamined skull characters). Although tines" (Cadle, 1984c, p. 643). problematicus becomes one of the smallest COMPARISONS: Externally, Liophis proble- species ofLiophis and has slightly fewer sub- maticus is quite different from its presump- tive intragroup relatives, which are mostly much larger snakes. The species of Erythro- 7 I suspect that the situation with Liophis may prove lamprus are brightly ringed coral snake mim- similar to that ofanother large assemblage ofNeotropical ics, whereas the usually crossbanded, heavy- snakes-Rhadinaea-for which the last reviser (Myers, bodied Xenodon and resemble 1974) was reasonably able to demonstrate the existence Waglerophis ofeight natural (monophyletic) species groups but could terrestrial vipers. The species of Lystrophis express only a degree of dubious faith that they were all are crossbanded or blotched and have an up- congeneric. Indeed, faith was not enough. New as well turned transversely keeled rostrum that is as reinterpreted morphological evidence (Myers, unpub- supported above by a longitudinal median lished), and supportive albumin immunology (Cadle, keel. Umbrivaga is characterized by a low 1984b), will necessitate a partitioning of Rhadinaea. 8 AMERICAN MUSEUM NOVITATES NO. 2853 caudals (36 vs. 38-106) than recorded by tional specimens not available to Roze, in- Dixon (1980, p. 22), only the questionable cluding two in the collection ofthe University rostral keel might add a structural character of Michigan Museum of Zoology and two at not already present in the genus (but see fn. the American Museum. Especially significant 3). All things considered, the allocation of is AMNH 117671 from Rancho Grande, problematicus to Liophis (s.l.) seems to me which has an everted hemipenis. In addition, the best taxonomic solution at this time. It notes on another (living) specimen from Ran- would be reassuring to identify some appar- cho Grande were provided by the late Scott ently closely related congener but I am unable J. Maness (in litt., Dec. 2, 1974).8 to do so. Aspects ofthe color pattern ofprob- HEMIPENIS: The left uneverted organ of lematicus are distinctive among Liophis, few UMMZ 124221 extends in situ to the end of ofwhich show any tendency toward pale nape subcaudal 8, being bifurcated from the base spots or pale labial markings. The head pat- of subcaudal 7 and with the two slips of the tern ofLiophis problematicus is more similar major retractor muscle merging at the end of to some species ofRhadinaea (e.g., compare subcaudal 9; the major retractor muscle orig- fig. 3 with Myers, 1974, fig. 26) than to any inates at the level ofsubcaudal 24. The sulcus Liophis. Most species of Liophis have the spermaticus proximally lies on the lateral wall upper lip basically pale, although some have and divides below the middle of the hemi- dusky labials owing to a suffusion of mela- penis, with the branches ending apically in nophores; only a few species have dark su- nude areas of longitudinally folded tissue. pralabials with definite pale markings, which (Dixon [1983, p. 128] unintentionally im- usually take the form of a horizontal stripe plied that the hemipenis is nonlobed when or line, in contrast to the oblique markings he described it as having "a" smooth apical in problematicus. One such species charac- disc.) terized by a white labial line is Liophis wil- That the distal nude areas of the retracted liamsi, which, like Liophis problematicus, is hemipenis do form a slightly depressed apical a small snake that might also be (and has disc on each lobe is proved by an everted been) confused with Rhadinaea. Even though organ of AMNH 117671 (fig. 7). The sulcus it is not closely related to L. problematicus, spermaticus has an essentially centrifugal ori- so little is known about L. williamsi that I entation, which is to say that the branches of take this opportunity to make available the the sulcus diverge widely and enter onto the following data. apical discs almost from opposite sides ofthe hemipenis. As is typical of bilobed hemi- NOTES ON LIOPHIS WILLIAMSI penes with nude apical discs, that of Liophis williamsi is noncapitate and is acalyculate This species was described by Roze (1958) (contra Roze), being ornamented solely with as Urotheca williamsi on the basis of four spines. specimens collected in 1929 and 1949, in COLOR IN LIFE: According to Maness, a liv- montane cloud forest of the Cordillera de la ing adult, about 495 mm in total length, had Costa, northern Venezuela. Roze misinter- a clear whitish cream line (on a dark upper preted the structure of the hemipenis, which lip) extending under the eye and posteriorly he thought was apically calyculate. I pointed for a distance ofabout 1.2 cm onto the neck; out the fact ofthe misinterpretation, without eye with a light orange iris and a round pupil; giving details, in order to remove williamsi tongue tips black. Dorsum light brownish ol- from consideration in the Rhadinaea- Uro- ive, with yellowish skin between the scales; theca complex (Myers, 1974, p. 21). Dixon four black lines starting posteriorly on body, (1980, pp. 17, 25; 1983) later associated the each about 1 /2 scales wide and the two dorsal name williamsi with Liophis. Cadle (1984a) lines merging to one on tail. Venter anteriorly was able to obtain a sample ofserum albumin and included the species in his study of immunological distances among South Ameri- 8 Mr. Maness was at that time a Peace Corps Volunteer can xenodontines. in Venezuela. He and a companion died tragically in I have seen the holotype, and four addi- 1981, while fighting a fire in Florida scrubland. 1986 MYERS: NEW SNAKE FROM ANDES 9

was found alive on a road at about 10 a.m., and the live specimen described by Maness was found resting in a small patch ofsunlight at midday (12 p.m.). Maness noted that one was very passive, never attempting to bite and that the cloacal secretion had a "not so unpleasant odor." He also observed that it was "a very strong con- strictor, when wrapped around a finger or thumb, cutting off circulation within min- utes." Maness wrote that although this snake was passive, "on occasion it would flatten the neck some 8 cm in length from behind the head, as seen in Leimadophis9 [and] Umbri- vaga." Such neck spreading would promi- nently display the yellow skin between the scales of L. williamsi. The defensive neck spreading just described for Liophis williamsi is of apparent significance in the suprageneric scheme ofre- Fig. 7. Hemipenis ofLiophis williamsi (AMNH lationships, as discussed below. 1 17671). Everted right organ, sulcate side, x 6.8. A BEHAVIORAL SYNAPOMORPHY light creamy yellow with indistinct small black Horizontal (dorsoventral) neck flattening or grayish spots or dots (venter apparently behavior (fig. 8) is widespread in the Xeno- becoming gray posteriorly [Maness's notes not dontini. The display is elicited in newly cap- explicit]), and uniform bright yellow under tured snakes that have been restrained from tail as in sympatric Rhadinaea multilineata. escaping but which are allowed some free- Brief color notes accompanying AMNH dom of movement while being prodded or 1 17671 indicate a similarly colored specimen otherwise annoyed (beware that there may be with white labial line, yellowish skin between no correlation between this behavior and scales, and venter "grey-yellowish." A third presence or absence ofdefensive biting). Lio- specimen (UMMZ 124224) was said by Test phis williamsi, L. zweifeli, and Umbrivaga et al. (1966, p. 44) to have had the belly "gray [mertensi] are referred to above; Vanzolini et with pinkish flush and the interstripe areas al. (1980, p. 60) described defensive neck posteriorly a pale brown"; I would add that spreading for the monotypic genus Waglero- the specimen mentioned by Test et al. is a phis; and my own fieldnotes record the be- juvenile of 210 mm total length and that (in havior for species as diverse as Erythrolam- preservative) the black head color extends for prus bizona, Liophis epinephelus, and a distance of five scales onto the nape and Xenodon rabdocephalus. Other species, es- there are poorly defined black crossbars on pecially of Liophis (s.l.), can be added from the anterior one-third of the body-both the scattered literature references; J. R. Dixon (in head-cap and crossbars being pattern features litt., Feb. 12, 1986) has witnessed the behav- that are absent or inconspicuous on adults. ior in the eight species of Liophis that he has See Dixon (1983) for more detailed descrip- examined alive. Lystrophis is the only genus tion of pattern in preserved specimens. of Xenodontini for which I am unaware of BEHAVIOR: Test et al. indicated that wil- an observation of neck flattening explicitly, liamsi is a rare snake at Rancho Grande; they found two dead on a road and a third one 9 Maness probably was referring to Leimadophis under dead leaves on a trail. But it is diurnal, (=Liophis) zweifeli, which occurs at Rancho Grande. or at least partly so, and not a nocturnal Test et al. (1966, p. 42) described the neck flattening species as they suspected. AMNH 117671 behavior of this species. 10 AMERICAN MUSEUM NOVITATES NO. 2853

W-4r, -W'W -

Fig. 8. Neck flattening by Liophis epinephelus (El Valle de Anton, Panama), a defensive threat display that is widespread in the Xenodontini. This individual also has inflated the fore part of its body just behind the horizontally spread neck; it made no attempt to bite, eventually abandoning the display and hiding its head under its body. although the whole body can be flattened in constituting the colubrid subfamily Xeno- an evidently complex behavioral repertoire dontinae, with notable exceptions including (Greene, 1973, p. 151). Heterodon, Hydrodynastes, and the tribe Xe- Horizontal neck spreading, or "hood" dis- nodontini. Therefore, in addition to hemi- play, has evolved more than once and is tax- penial morphology and albumin correspon- onomically and geographically widespread dence, neck flattening appears to provide a among snakes, as has long been recognized behavioral trait that is consistent with rec- (Noble, 1 92 1). Strictly defined, however, this ognition of the Xenodontini as a monophy- behavior is relatively less common among letic group -even ifthe behavior should prove snakes generally than was suggested by Car- to be lacking in some species, as might be penter (in Carpenter and Ferguson, 1977). expected in such a diverse assemblage of Basing his conclusions on an extensive search snakes. For example, Brazilian Erythrolam- ofthe literature, which is difficult to analyze, prus aesculapii possibly does not display ac- Carpenter coded neck-flattening (hood) be- cording to P. E. Vanzolini (verbal commun.), havior in species representing 35 of 94 listed who pointedly emphasized, however, that genera ofColubridae (containing roughly 300 prudent collectors do not give much oppor- genera). But this was an inappropriate des- tunity for freedom of expression to newly ignation for some snakes, such as those that captured corals and their mimics unless they properly were also coded for flattening the are specifically looking for defensive behav- entire trunk (e.g., the xenodontine genus Ni- ior.'0 I noted that a Panamanian E. bizona nia and New World natricines of the genera '0Added in press: Vanzolini subsequently wrote (in Clonophis, Nerodia (as Natrix), Storeria, litt., March 17, 1986) that, "I have spent the last half Thamnophis, and Tropidoclonion). hour teasing an Erythrolamprus [a.] venustissimus. A In any event, neck flattening is an uncom- healthy and cooperative snake, very willing to strike. No mon behavior among the nearly 100 genera flattening." 1986 MYERS: NEW SNAKE FROM ANDES 11 would "conspicuously flatten about a third spreading, or hood display, would seem to of its neck when disturbed" (fieldnotes for have been separately derived in these three Univ. Kansas 110693); R. G. Zweifel simi- lineages of New World snakes. larly observed another Panamanian bizona that "flattened out the neck somewhat in the fashion ofHeterodon" (fieldnotes for AMNH ACKNOWLEDGMENTS 90019); H. G. Greene (in litt., Jan. 22, 1986) I thank curators Robert F. Inger and Hy- saw neck flattening in one E. bizona and three men Marx, Field Museum ofNatural History E. mimus in Costa Rica and described a pho- (FMNH), for permission to study the new tograph (by R. W. McDiarmid) of a Costa snake. An important specimen of Conophis Rican E. mimus that was simultaneously dis- (fn. 4) was lent by Dr. Arnold G. Kluge, Uni- playing a coiled tail while elevating a flat- versity of Michigan Museum of Zoology tened neck. Thus, observations on Central (UMMZ). Figure 6 was drawn by Frances L. American bizona and mimus suggest that Waite (Gibson). For reading and comment- possible absence of such behavior in other ing on the manuscript, I am grateful to Drs. Erythrolamprus could be due to evolutionary John E. Cadle, James R. Dixon, Harry W. loss. Attrition of synapomorphies during Greene, Samuel B. McDowell, and Paulo E. adaptive radiation is a major hindrance in Vanzolini. phylogenetic analysis, and the problem becomes more serious when homoplasy is in- volved. LITERATURE CITED Although I suggest above that neck flattening is a behavioral synapomorphy ofthe Xe- Cadle, John E. nodontini, there is an alternative hypothesis 1984a. Molecular systematics of Neotropical that might be kept in mind: Notwithstanding xenodontine snakes: I. South American that North American Heterodon and South xenodontines. Herpetologica, vol. 40, no. 1, pp. 8-20. American Hydrodynastes are excluded be- 1984b. [Same title]: II. Central American xe- cause offundamental hemipenial differences, nodontines. Ibid., pp. 21-30. the possibility that one or the other is a sister 1984c. [Same title]: III. Overview of xenodon- group ofthe Xenodontini has not been prop- tine phylogeny and the history of New erly tested by morphological criteria, al- World snakes. Copeia, 1984, no. 3, pp. though such relationships are contraindicat- 641-652. ed by genetic distance data (see Historical Carpenter, Charles C., and Gary W. Ferguson Summary under Discussion). If such a rela- 1977. Variation and evolution of stereotyped tionship could be corroborated, then neck behavior in reptiles. In C. Gans and D. spreading would have the appearance ofbeing W. Tinkle (eds.), Biology ofthe Reptilia, vol. 7. London, Academic Press, pp. synapomorphic for the larger group, hence 335-554. symplesiomorphic for the Xenodontini. Be- Cope, Edward Drinker havioral synapomorphy would at least re- 1894. The classification ofsnakes. Amer. Nat., quire the fewest assumptions ifa sister-group vol. 28, pp. 831-844, pls. 27, 28. relationship were established on the basis of 1895. The classification ofthe Ophidia. Trans. derived anatomical characters. Unless there Amer. Phil. Soc., vol. 18, pt. 2, pp. 186- prove to be morphological differences in- 219 + pls. 14-33. volved in the neck spreading itself (a likeli- Dixon, James R. hood worthy ofinvestigation), homoplasy in 1980. The Neotropical colubrid snake genus behavior would be difficult if not impossible Liophis. The generic concept. Milwau- if kee Public Mus., Contrib. Biol. and to demonstrate either Heterodon or Hy- Geol., no. 31, pp. 1-40. drodynastes were postulated as the sister 1983. Systematics of Liophis reginae and L. group of the Xenodontini. But homoplasy is williamsi (Serpentes, Colubridae), with a necessary assumption if the groups are a description ofa new species. Ann. Car- judged as not related: Based solely on rela- negie Mus., vol. 52, art. 6, pp. I 13-138. tionships suggested by available immunolog- Dowling, Hemdon G. ical distance data (Cadle, 1 984a, 1 984c), neck 1969. The hemipenis of Philodryas Gunther: 12 AMERICAN MUSEUM NOVITATES NO. 2853

a correction (Serpentes, Colubridae). amphisbaenians. Jour. Herpetology, vol. Amer. Mus. Novitates, no. 2375, pp. 1- 7, no. 3, pp. 143-161. 6. Myers, Charles W. Dowling, Herndon G., and William E. Duellman 1974. The systematics of Rhadinaea (Colu- "1974-1978" [1978].11 Systematic herpetolo- bridae), a genus of New World snakes. gy: a synopsis offamilies and higher cat- Bull. Amer. Mus. Nat. Hist., vol. 153, egories. New York, privately printed by art. 1, pp. 1-262. senior author (as "Hiss Publications, Myers, Charles W., and Jonathan A. Campbell Publ. Herpetology, no. 7, with minor 1981. A new genus and species of colubrid revisions, 1975-1978"), issued loose- snake from the Sierra Madre del Sur of leaf, [302 pp.]. Guerrero, Mexico. Amer. Mus. Novi- Dowling, Herndon G., and Frances W. Gibson tates, no. 2708, pp. 1-20. 1970. Relationships ofthe Neotropical snakes Neill, Wilfred T. Hydrodynastes bicinctus and Cyclagras 1964. The phylogenetic position ofDugandia gigas. Herpetol. Rev., vol. 2, no. 2, pp. bicincta (Serpentes: Colubridae). Her- 37-38. petologica, vol. 20, no. 3, pp. 194-197. Dunn, Emmett Reid Noble, G. Kingsley 1928. A tentative key and arrangement of the 1921. Snakes that inflate. Nat. Hist., vol. 21, American genera of Colubridae. Bull. no. 2, pp. 166-171. Antivenin Inst. Amer., vol. 2, no. 1, pp. Roze, Janis A. 18-24. 1958. A new species of the genus Urotheca Greene, Harry W. (Serpentes: Colubridae) from Venezue- 1973. Defensive tail display by snakes and la. Breviora, Mus. Comp. Zool., no. 88, pp. 1-5. Test, Frederick H., Owen J. Sexton, and Harold Heatwole 11 Bibliographic note: This was cited several times in 1966. Reptiles of Rancho Grande and vicin- 1975 with varying pagination and a 1974 publication ity, Estado Aragua, Venezuela. Misc. date, which has caused some confusion (to H. M. Smith, Publ. Mus. Zool., Univ. Michigan, no. for example, 1977, Jour. Herpetology, vol. 1 1, p. 115), 128, pp. 1-63. but the present version seems to be the first one actually Thompson, J. C. distributed or made available for purchase. W. E. Duell- 1913. Oxyrhopus trigeminus Dumeril and Bi- man (in litt., Sept. 5, 1985) determined the publication bron the type of Erythroxyrhopus gen. date to be December 1978. nov. Proc. Acad. Nat. Sci. Philadelphia, Dowling et al. (1983, Jour. Zool., London, vol. 201, vol. 65, pp. 78-80. p. 327) recently departed from the usual format by citing Underwood, Garth Dowling and Duellman as editors ofthe work, but Dow- 1967. A contribution to the classification of ling and Duellman agree in their preface that "the work snakes. London, British Museum (Nat- should be considered a joint effort." The "major con- ural History), publ. no. 653, x + 179 tributor" was supposed to have been indicated by the PP. "initials and date at the end of each section," although Vanzolini, P. E., Ana Maria M. Ramos-Costa, and this was done in only one instance (WED on p. 7.2); but Laurie J. Vitt there is no suggestion of other authors. 1980. Repteis das caatingas. Rio de Janeiro, This is a useful compendium. Users should realize, Acad. Brasileira Cien., [vi] + 161 pp. + however, that all or most ofthe approximately six dozen 40 pls. "AMNH drawings" have been previously published, al- Wellman, John though the original authors (e.g., Bogert, Noble) are not 1963. A revision of snakes of the genus Co- cited; a few names accompanying these drawings have nophis (family Colubridae, from Mid- not been in use for years, having been simply copied dle America). Univ. Kansas Publ. Mus. from the old original artwork. Nat. Hist., vol. 15, no. 6, pp. 251-295.

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