Rare Snakes—Five New Species from Eastern Panama: Reviews Of

Home , Geophis, Psomophis, Trimetopon

PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, NY 10024 Number 3391, 47 pp., 20 ®gures, 2 maps, 3 tables February 19, 2003

Rare SnakesÐFive New Species from Eastern Panama: Reviews of Northern Atractus and Southern Geophis (Colubridae: Dipsadinae)

CHARLES W. MYERS1

CONTENTS Abstract ...... 2 Resumen ...... 2 Introduction ...... 3 Gorgas Memorial Laboratory and the Panama Snake Census ...... 4 Methods of Study ...... 5 Genus Atractus ...... 8 Key to Panamanian Atractus ...... 9 Atractus clarki Dunn and Bailey ...... 10 Atractus darienensis, new species ...... 16 Atractus depressiocellus, new species ...... 20 Atractus hostilitractus, new species ...... 22 Atractus imperfectus, new species ...... 25 Genus Geophis ...... 28 Key to Panamanian Geophis ...... 28 Records Excluded from Central Panama ...... 29 Geophis bellus, new species ...... 30 Geophis brachycephalus (Cope) ...... 37 Geophis, species inquirenda ...... 38 Geophis hoffmanni (W. Peters) ...... 40 Museum Abbreviations and Acknowledgments ...... 43 References ...... 44

1 Curator Emeritus, Division of Vertebrate Zoology (Herpetology), American Museum of Natural History.

ABSTRACT The South American genus Atractus barely enters political North America on the eastern half of the Isthmus of Panama, where it is extraordinarily rare. Collected over a period of 39 years, the ®ve Panamanian specimens of Atractus known to the author represent ®ve species! Four new species are described: A. darienensis, A. hostilitractus, A. imperfectus, and A. depressiocellus. The ®fth species is A. clarki Dunn and Bailey, for which a second specimen is reported from the Colombia ChocoÂ. The noncapitate hemipenis of A. clarki may be primitive in being calyculate and deeply bilobed. The morphologically convergent Geophis is primarily a Middle American genusÐMexico to western Panama, with two or three outlying species in the western Andes of Colombia (G. betaniensis, G. nigroalbus, and probably G. hoffmanni). The genus is unrecorded from eastern Panama, and a few old records for central Panama seem to have been based on erroneous specimen data. Nonetheless, the genus does occur in east-central Panama, based on two specimens of G. hoffmanni (W. Peters) and on a specimen each of Geophis bellus, new species, and G. brachycephalus (Cope)Ðthe latter representing a disjunct population separated by about 340 km from those in the Boquete area of western Panama. Geophis bellus is a tiny snake differing from sympatric G. brachycephalus and South Amer- ican G. nigroalbus in characters of size, color, and hemipenis. Geophis brachycephalus may be a composite species in western Panama. Unicolored specimens from the Atlantic versant seem to differ from those in the polymorphic Boquete population in hemipenial and other characters, and they are set aside as a species inquirenda. The ®rst specimen of Geophis hoffmanni is reported from Colombia, but it lacks precise data. Atractus depressiocellus, A. imperfectus, Geophis bellus, G. brachycephalus, and G. hoffmanni are at least broadly sympatric on the ``Piedras-Pacora Ridge''Ðthe continental divideÐ between the upper drainages of the RõÂo Chagres and RõÂo Pacora, some 30 km northeast of Panama City. This relatively low upland likely is a premontane forest refuge, where some very rare snakes may be making a last stand prior to extinction.

RESUMEN El geÂnero Atractus esta ampliamente distribuido en SudameÂrica y tiene su limite norte en la regioÂn oriental del Istmo de PanamaÂ, adonde es extraordinariamente raro. ¡Los 5 uÂnicos ejemplares panamenÄos conocidos y estudiados por el autor, colectados a lo largo de 39 anÄos, corresponden a 5 especies! Cuatro son especies nuevas que se describen en este trabajo: A. darienensis, A. hostilitractus, A. imperfectus,yA. depressiocellus. La quinta especie es A. clarki Dunn & Bailey, de la cual se reporta un segundo ejemplar del ChocoÂ colombiano. Los hemipenes no capitados de A. clarki podrõÂan ser primitivos por ser caliculados y profunda- mente bilobulados. Geophis, que es morfoloÂgicamente convergente con Atractus, es primariamente un geÂnero mesoamericanoÐse distribuye desde MeÂxico hasta el oeste de PanamaÂ, con dos o tres especies aisladas en los Andes occidentales de Colombia (G. betaniensis, G. nigroalbus, y probablemente G. hoffmanni). Geophis no ha sido registrado hasta ahora en el este de PanamaÂ, y unos pocos registros viejos del centro de PanamaÂ parecen haber sido basados en especõÂmenes de procedencia incierta. Sin embargo, el geÂnero ocurre en el centro-este de PanamaÂ, como lo indican dos especõÂmenes de Geophis hoffmanni (W. Peters), un espeÂcimen de G. bellus (especie nueva), y un espeÂcimen de G. brachycephalus. Este uÂltimo representa una poblacioÂn disyunta, distante aproximadamente 340 km de las poblaciones mas cercanas en el oeste de PanamaÂ. Geophis bellus es una pequenÄita culebra que se diferencia de la especie simpatrida brachycephalus y de la sudamericana nigroalbus en caracteres de tamanÄo, coloracioÂn y hemipenes. La especie polimoÂr®ca G. brachycephalus podrõÂa ser una especie compuesta en el oeste de PanamaÂ. Ejemplares monocromaÂticos de la pendiente atlaÂntica parecen diferir en caracterõÂsticas de los hemipenes y de otros caracteres, y son mantenidos como especie inquirenda. Se reporta el primer espeÂcimen de Geophis hoffmanni de Colombia, aunque faltan datos precisos. Atractus depressiocellus, A. imperfectus, Geophis bellus, G. brachycephalus,yG. hoffmanni son ampliamente simpatridas en la ``cresta Piedras-Pacora'', la divisoria de los altos drenajes de los rõÂos Chagres y Pacora, unos 30 km al nordeste de la ciudad de PanamaÂ. Esta serranõÂa 2003 MYERS: SNAKES FROM PANAMA 3

baja es probablemente un refugio de bosque montano, adonde algunas serpientes muy raras podrõÂan presentar una uÂltima resistencia a la extincioÂn.

INTRODUCTION albiceps is Barro Colorado Island, which This paper is about some rare snakes on supports one of the world's best known trop- the eastern half of the Isthmus of Panama. ical herpetofaunas (Myers and Rand, 1969; Biologists familiar with the great diversity of Rand and Myers, 1990). Perhaps Tantilla al- serpents in tropical rain forests know that rar- biceps is extinct, at least on Barro Colorado ity or the appearance of rarity is compounded Island. Perhaps biologists on the island have by several factors, including: (1) many, in- failed to differentiate an occasional specimen deed most, wet-forest snakes seem to have from the more common and similar appear- low population densities relative to temperate ing Enuliophis sclateri. Such explanations species; (2) many are hard to ®nd because of are unveri®able and unsatisfactory. secretive habits; and (3) some are less likely Five or six species in the two genera of to be encountered because of small geo- concern in this paper seem to be, like the graphic ranges and/or specialized microhab- example above, really rare, although to be itats. sure they come from less well-trodden With time, however, suf®cient specimens ground. The terrestrial or semifossorial accumulate to allow approximation of geo- snakes of the genus Atractus reach the north- graphic and ecologic distributions and as- ern limits of the generic range in east-central sessment of morphological variation. But Panama. To my knowledge, only ®ve Pana- there are exceptions to this generalizationÐ manian specimens of Atractus have been col- there are snakes seemingly rare in an abso- lectedÐin the years 1936, 1938, 1966, 1967, lute sense, so rare as to resist veri®able ex- 1974Ðwith each specimen a different spe- planation of their rareness. For example, the cies, of which one was named in 1939 and little Tantilla albiceps is known to science the other four are described herein. only from a single specimen that was found The second genus of concern, the semi- in central Panama three-quarters of a century fossorial Geophis, ranges throughout much ago (Barbour, 1925)2; the type locality of T. of Middle America and occurs also in the western Andes of northern South America. 2 PeÂrez-Santos and MartõÂnez (1997: 451) erroneously Geophis is well known from Costa Rica and reported Tantilla albiceps from Isla Coiba off the Paci®c western Panama, but published records for coast of western Panama; this record is based on a specimen of the widespread Enulius ¯avitorques. The col- central Panama are based on erroneous lo- lector, Ignacio De la Riva, kindly provided photographs cality data and the genus has not been doc- of the specimen in life, showing a smooth-scaled, long- umented from the eastern half of the isthmus. tailed small snake, dark brown above, grayish white on I am still unaware of specimens from ex- the lower few scale rows. The dark dorsal coloring is treme eastern Panama, but four snakes rep- conspicuously broken by a yellowish nuchal collar crossing the rear halves of the parietals; the collar is resenting three species of Geophis were col- interrupted by a small irregular brown blotch posteriorly lected in east-central Panama in the 1950s on the interparietal suture. The left side of the head in and 1960s. Two specimens represent a sim- pro®le shows a prominent rostral, a longer-than-high lo- ple range extension of the widely distributed real plate entering the eye (no preocular), two postoculars, and 1 ϩ 2 temporals. All these characters match G. hoffmanni, whereas a third specimen pro- well with E. ¯avitorques. vides a widely disjunctive eastward exten- According to J.M. Savage (personal commun.), the sion of G. brachycephalus. The fourth spec- unpublished notes of E.R. Dunn contain reference to a imen is of an undescribed species that is second specimen of Tantilla albiceps from Barro Colo- rado obtained in 1952, but no one else seems to have named herein. been aware of the specimen, and its whereabouts is un- Apparent rarity often implies insuf®cient known (there is no second specimen of T. albiceps en- collecting, and certainly the survey of the tered in the electronic databases of ANSP, FMNH, or rich Panamanian herpetofauna is not (and MCZÐthe main depositories for Panamanian specimens examined by Dunn in the early 1950s). No other spec- may never be) complete. Nonetheless, Pan- imens have been reported (IbaÂnÄez et al., ``1995'' [1997]: amanian snakes have been better collected 153, 156). than in most tropical countries. My claims of 4 AMERICAN MUSEUM NOVITATES NO. 3391 unusual rarity for some of the snakes in hand Dunn and Bailey, 1939).3 Dunn obtained can be put in perspective by consideration of background information on the census meth- the activities of the Gorgas Memorial Labo- ods during his visits to GML starting in ratory (GML), which operated out of Panama 1939, and his 1949a paper is the most com- City from its opening in 1929 until its un- plete and easily accessible summary of fortunate demise in 1991. (GML's parent cor- Clark's GML snake census. poration in Washington, D.C.Ðthe Gorgas Dunn's 1949a account, however, did not Memorial Institute of Tropical and Preven- include the census collections from most of tive Medicine, Inc.Ðsubsequently was reor- western Panama nor from the limited uplands ganized in name under the Ministry of of eastern Panama (see below). But new spe- Health in Panama, but GML had been de- cies, new country records, and other note- pendent on a permanent annual contribution worthy specimens from those areas were re- from the U.S. Congress for its core support.) ported elsewhere (Dunn, 1942; Dunn and Bailey, 1939). Unfortunately, over 90% of GORGAS MEMORIAL LABORATORY (GML) the GML specimens were apparently discard- AND THE PANAMANIAN SNAKE CENSUS ed after Dunn's study, except for whole specimens and selected heads especially of rare Dr. Herbert C. Clark (1877±1960), ®rst Di- species that were placed in major museums. rector of the Gorgas Memorial Laboratory, Fewer than 1000 specimens with data were initiated a snake census in Panama and con- saved from the Panama snake census, which ducted it at varying levels of intensity from spanned a quarter of a century.4 January 1929 through 1953 (he retired in A total of 13,745 snakes, mostly heads 1954). Not a new concept, the census was only, were accumulated in the overall period started as an extension of a then ongoing 1929±1953 (Wright, l970: 269). Dunn Central American snake census that used the labor forces on plantations of the United 3 In describing one of the heads as a new species (Dip- Fruit Company. Clark had earlier supervised sas nicholsi [see Cadle and Myers, 2003]), Dunn (1933) the snake census in Tela, Honduras, but, with acknowledged ``the authorities of the Museum of Com- parative Zoology for permission to examine the collec- the resources of GML, Clark expanded the tion'' (Dunn, 1933: 193). Later, however, it seems cu- census in Panama well beyond the planta- riously inappropriate for Dunn (1949a: 39, 55) to have tions and continued it for a quarter of a cen- stated that the specimens were sent directly to him ``dur- tury. ing the years 1933 to 1945 inclusive, by Dr. H. C. Clark's primary aim was to learn the ven- Clark''Ðwithout mention of the MCZ or, especially, without acknowledgment to the late Thomas Barbour omous species in Panama and to assess their (died in 1946), who funded Dunn's early ®eldwork in relative medical importance (Clark, 1942). the Neotropics (Adler, 1989: 93). According to the GML Bounties were paid for killed snakes (mostly annual reports, yearly shipments of snakes were sent to heads), which were accumulated in formalin the MCZ at least through 1938 and 1939, and the MCZ continued to be mentioned as a cooperating institution in central locations around the country. The through 1943; Dunn's identi®cation services were ac- GML snake census was initiated in cooper- knowledged starting in the report for 1939, and the MCZ ation with the old Antivenin Institute of is not mentioned after 1943. America, one of whose founders was Thom- 4 The following counts are derived from collection da- as Barbour. In 1932, Barbour involved the tabases: Clark sent the early material to the MCZ, including 411 specimens from the Panama census and Museum of Comparative Zoology (MCZ) of some 300 from his prior Honduran census. E.R. Dunn Harvard University, which ``became actively subsequently presented 273 of Clark's Panamanian spec- interested in the studies and supported them imens to the ANSP, starting with a few specimens in in part'' (Wright, 1970: 269). Early ship- 1939 and with the bulk being given during the early 1940s. Finally, with Clark's approval, Dunn and GML ments were sent to the MCZ (e.g., Clark, staff member H. Trapido (see footnote 20) saw to it that 1937: 12), from where Barbour turned over 256 specimens from the Panama census went to the the identi®cation and study of the collection FMNH after Dunn's examination (Clark, 1952: 20). A to his former student E.R. Dunn, who iden- jar of uncataloged heads without data is also at FMNH (Alan Resetar, personal commun.). In Panama, a small ti®ed most of the material. Dunn eventually number of specimens were kept for identi®cation pur- took over the study and reported on the re- poses at least at GML and at Barro Colorado Island sults (Dunn, 1933, 1942, 1949a, 1949b; (Myers and Rand, 1969: 5). 2003 MYERS: SNAKES FROM PANAMA 5

(1949a: 39, 55) reported on nearly 80% of herein; the other Geophis specimens repre- the collection, that is on ``10,960''5 speci- sent two species named in the 19th century. mens from the lowlands of central Panama The ®rst two of the aforesaid nine speci- (``Chagres collection'') and from the Paci®c mens were obtained by the GML census in lowlands of west-central and eastern Panama 1936 and 1938. I collected another two while (combined collections from ``CocleÂ-Herre- resident Visiting Scientist at GML during ra'', ``Sabanas'', ``DarieÂn''). Excluding 1073 1964±1967. And ®ve experienced ®eld men museum specimens, savanna collections, and (several earlier trained as collectors at GML) those collections from west of the Panama each caught one of the other ®ve specimens Canal, Dunn's (1949a; table 7) list of 70 spe- in the 23-year period 1952±1974. cies includes 2500 snakes from forested low- As a further indication of rarity, it may be lands in the Chagres drainage of central Pan- noted that ®ve of the eight species (A. de- ama and 3044 snakes from areas of mixed pressiocellus, A. imperfectus, G. bellus, G. banana plantations and forest in the DarieÂn brachycephalus, G. hoffmanni) are from the of eastern Panama. region of the Piedras-Pacora Ridge, only Neither Atractus nor Geophis is included about 30 km northeast of Panama City. The in the nearly 11,000 lowland specimens ob- Piedras-Pacora RidgeÐa section of the contained by the GML snake census and report- tinental divideÐdelimits part of the south- 6 ed in Dunn's (1949a) summary paper. The eastern side of the Madden Lake watershed, absence of Atractus and Geophis from the and is now included in the Parque Nacional large lowland collections seems surprising at Chagres. A team of Panamanian herpetolo- ®rst blush, inasmuch as there are many gists (IbaÂnÄez et al., ``1994'' [1995]) surveyed Atractus occurring in the South American the herpetofauna of the Piedras-Pacora Ridge lowlands and a specimen of Geophis hoff- in 1990±1995; they invested 1253 man-hours manni was collected not far from the Panama of collecting in both wet and dry seasons (in- Canal in 1968. Recent sampling in the cen- cluding 166 man-hours at night in the wet season). Their list is extensive, totaling 131 tral Panamanian lowlands has yielded neither species of amphibians and reptiles, including Atractus nor additional specimens of Geo- 37 species of snakes. Among the small phis (IbaÂnÄez et al., ``1995'' [1997]). snakes collected was the third known speci- Clark also pursued collections of snakes men of Coniophanes joanae, but no new from upland areas, which are of limited ex- specimens of the still elusive Atractus or tent in central and eastern Panama. Dunn and Geophis. Bailey (1939: 3) reported on ``268 snakes The Piedras-Pacora Ridge may be a pre- from the uplands of eastern Panama gathered montane forest refuge for some small part of in 1936 to 1938 through the initiative of Dr. the fauna that is barely surviving the climat- H. C. Clark''. There are no Geophis in this ic-vegetational vicissitudes of the Pleisto- collection, which, however, provided the ®rst cene. The mingling of several very rare spe- two specimens of Atractus from Central cies in one place, including three species America. known only from that place, leads me to fear There are eight species to be reported in that it may be their ``last stand'' on the way the following pages (maps 1, 2). These eight to extinction. Should this be so, their Latin species are represented by a total of nine cen- names may be their only epitaphs. tral and eastern Panamanian specimensÐ®ve specimens of Atractus and four of Geophis. METHODS OF STUDY One Atractus was named in 1939; four specimens of Atractus and one of Geophis rep- Total length and tail length were determined by moderately stretching preserved resent ®ve ``new'' species that are named specimens along a metric rule. Head and 5 Apparently a minor error, repeated later (Dunn, snout lengths, eye length, and head plates 1949b). Adding the column totals in Dunn's (1949a) ta- and their sutures were measured to the near- ble 7 gives a grand total of 10,693 specimens. His count of 1073 museum specimens on page 55 agrees with the 6 Madden Lake is shown as Lago Alajuela on recent column totals in the table. maps. 6 AMERICAN MUSEUM NOVITATES NO. 3391

Map 1. Eastern Panama, showing locality records for the ®ve Central American species of Atractus. One species (A. clarki) is also known from a specimen of inde®nite locality in the Colombian ChocoÂ, the others are known only from the Panamanian holotypes. Type localities: (1) Santa Cruz de Cana, an old gold mine; (2) north end of the SerranõÂa de Pirre; (3) ``Morti Hydro'', a temporary helipad near the RõÂo MortõÂ; (4) Piedras-Pacora Ridge; (5) Cerro Jefe on the Piedras-Pacora Ridge. est 0.1 mm with an ocular micrometer in a whereas one such as eye length/eye-to-lip dissecting microscope. Some measurements may or may not match a ®gure, depending are used for comparisons, but most were con- on perspective of the drawing (e.g., dorso- verted to proportions for purposes of descrip- lateral or lateral) and degree to which the lip tions. It bears emphasizing that snout length is ¯ared. (tip of snout to eye) was not measured on the The drawings are my own, made with the oblique angle but on the sagittal plane by oc- aid of a camera lucida ®tted to a Wild dis- ular micrometer, whereas head plates and the secting microscope. I photographed speci- distance from eye to lip were determined mens positioned under glass on a raised plas- along the plane of the greatest dimension, tic platform to eliminate shadows, and under meaning that the head had to be differently alcohol to reduce glare from re¯ective sur- inclined for different measurements. There- faces of scales. fore anyone comparing the descriptions with Ventral plates were counted by the Dowl- the drawings will have to be alert to the usual ing (1951) method, starting with the ®rst problems of parallax: In no case, for exam- plate bordered on each side by the ®rst dorsal ple, will a proportion such as nasal plate/lo- scale rows; gular plates anterior to this, if real plate be retrievable from the ®gures, wider than long, were termed preventrals 2003 MYERS: SNAKES FROM PANAMA 7

Map 2. Locality records for three species of Geophis known to occur in eastern Panama. The easternmost localities indicate broad sympatry on the Piedras-Pacora Ridge in PanamaÂ Province.

(Myers, 1974: 37) for purposes of comparing method over the older one summarized by type specimens with old original descriptions Schmidt and Davis (1941: 26): in which these plates were included in ven- The ventrals and caudals of snakes correspond (with tral counts. some variation) to the number of vertebrae. It is cus- Dowling's method has the advantage of tomary to count their number beginning beneath the giving a standard starting place for ventral chin with the ®rst one distinctly wider than long [emphasis added], and excluding the anal plate . . . The counts that was said to correspond (in some ®rst caudal, when they are in two rows, is the ®rst snakes) in position with the atlas-axis artic- one that meets one of the opposite side. It is custom- ulation of the vertebral column. But the ®rst ary to count them on only one side, and to include ventral is not always quickly ascertained in the terminal single scale [however, most workers nowadays do not count the terminal spine as a sub- the case of small, poorly preserved snakes caudal]. with shiny scales, and I am sorry to have adopted this method some years ago. I sel- Downs (1967: 23) may have found a reason- dom notice more than a few ``preventrals'', able compromise between the above and but comparisons with older literature are less Dowling's sometimes tedious method of de- accurate if these are not counted and added termining the ®rst ventral: to the total. Therefore, I would not be critical The ventrals were counted from the ®rst scale clearly of anyone refusing to adopt the Dowling twice as broad as long . . . this system does not fol- 8 AMERICAN MUSEUM NOVITATES NO. 3391

low the suggestion of Dowling (1951), but in practice in Fairchild and Handley (1966), whose im- is identical or nearly identical with it. portant gazetteer of GML and other collect- The practical reason for the above digression ing stations was based on an earlier 3-sheet is to be seen in this paper in footnote 19. As 1:500,000 map of the Republic. a further aside, only a few workers (e.g., Pe- ters, 1960: 10±12) have mentioned the oc- GENUS ATRACTUS casional divided or ``half-ventrals'', which Atractus is a large South American genus may occur anywhere but are most common whose range is now known to include the in some snakes just anterior to the anal plate. eastern half of Panama (map 1), where it These correspond to anomalous vertebral du- seems to be exceedingly rare. The four whole plications (King, 1959) and are not counted, specimens (®gs. 1, 2) and the one head (®g. although it may be useful to mention their 10) known to me represent ®ve species! Two occurrence in holotypes. of these specimens come from the low up- Except when a dentigerous bone was dis- lands east of the Canal Zone and three are sected out for illustrating, maxillary teeth from DarieÂn; all localities are in well-drained were counted in situ (Myers, 1974: 27); the (100 to Ͼ500 m) monsoon rain forest (My- once common practice of excising maxillae ers, 1969). from rare snakes for the sole reason of count- I have been unable to match any of the ing teeth is neither necessary nor justi®ed. Panamanian specimens with previously Hemipenes were treated as described in named South American species. But identi- Myers (1974), except that everted organs fying Atractus is often dif®cult owing to the were in¯ated with carmine-dyed petroleum many inadequate, poorly illustrated descrip- jelly (rather than wax or latex). A few hem- tions that usually emphasize characters pre- ipenes in preserved specimens were manu- sent in a majority of species. A major review ally everted from either the retracted state of the genus is needed. There have been good (®g. 17) or from partially everted organs taxonomic and variational studies for Ecua- (®gs. 14, 20B), after they were dissected out dor (Savage, 1960), Venezuela (Roze, 1961, and then soaked ®rst in glycerin and ®nally 1966), Surinam (Hoogmoed, 1980), and east- in a saturated solution of trisodium phos- ern and central Amazonia (Cunha and Nas- phate to restore elasticity. Although hemi- cimento, 1983; Martins and Oliveira, penial eversions made from preserved spec- 1993)Ðbut the situation is less good for Co- imens may be fully everted with all struc- lombia, which has a rich Atractus fauna of tures showing, some such organs may be less several dozen named species, whose varia- completely in¯ated (i.e., less robust) than tion and relationships are virtually unknown fresh preparations. Techniques are reviewed (lists of nominal species in Daniel, 1949; Pe- by Myers and Cadle (MS). ters and Orejas-Miranda, 1970; Sanchez-C. All specimens were collected long before et al., 1987; PeÂrez-Santos and Moreno, the availability of Global Positioning System 1988). Although a majority of Colombian (GPS) satellite receivers. The detailed 1: Atractus are Andean, it would not be sur- 50,000 map series of Panama is still incom- prising if an older Colombian (or other South plete and, with one exception (see Atractus American) name were eventually found to be hostilitractus), topographic maps of this applicable to one of the purportedly new spe- scale are still unavailable for the type local- cies herein described as Panamanian endem- ities in this paper. Therefore, approximate ics. To ease the burden of workers who may geographic coordinates were determined have to decide this point without ready ac- mainly from the maps Canal Zone and Vi- cess to all holotypes, the ®ve Panamanian cinity 1:100,000 (1957 edition, U.S. Army specimens are described and illustrated in Map Service, Washington, DC) and the 12- some detail. I additionally discuss and illus- sheet Mapa General de la RepuÂblica de Pan- trate a Colombian specimen of Atractus clar- amaÂ 1:250,000 (1st ed., circa 1966±1967, ki Dunn and Bailey, a species heretofore re- DireccioÂn de CartografõÂa, Panama City). ported only from the Panamanian holotype. Some coordinates differ slightly from those This genus has been confused with Geo- 2003 MYERS: SNAKES FROM PANAMA 9 phis, but, in Panama, known Atractus are cially close relationship between Atractus easily distinguished by the generic character- and the Middle American Adelphicos. istic of a single pair of genials (two pairs in Panamanian Atractus share a number of southern Geophis) and 17 rows of smooth features7 that are possessed by a majority of scales (15 rows of scales, smooth or poste- species in the genus (e.g., see data matrix in riorly keeled in southern Geophis); also, the Peters and Orejas-Miranda, l970: 24±26). southern Geophis have a usually diamond- They have 17 dorsal scale rows, large pre- shaped or rhomboidal frontal plate that is frontals that are in contact with the eye, usu- strongly angular anteriorly (as in ®g. 12), ally two postoculars (2/3 in one), no preo- whereas Panamanian Atractus have a more cular, an elongated loreal plate, seven su- normal colubrid frontal that is roughly tri- pralabials, ®rst pair of infralabials in contact angular or pentagonal, with only a small an- behind the mental, and a basically cross- terior apex at the prefrontal suture (®g. 3). banded or blotched color pattern. The ®ve Unfortunately, the four whole specimens Panamanian specimens/species are identi®- of Panamanian Atractus are all females and able by the following key (see also compar- furnish no data on male genitalia, although ative ®gures, diagnoses, and table 1): opportunity is taken to describe and illustrate a hemipenis from the Colombian specimen KEY TO PANAMANIAN ATRACTUS of A. clarki. Hemipenial differences mentioned by Savage (1960: 30) have since bro- 1. Venter black or clouded with dark pigment; ken down with further study of Geophis (see no pale dashes on lower scale rows; eye Downs, 1967: 184, and discussion herein un- length less than or scarcely greater than der Geophis bellus), and the distinction is distance from its lower edge to lip .... 2 further blurred by the hemipenis of Atractus ± Venter pale (®g. 2A); lower scale rows with clarki (q.v.). Cadle (1984) considered Atrac- pale centers; eye length noticeably greater tus a South American member of a mainly than distance to lip (®g. 3A) ...... Central American xenodontine cladeÐnow ...... A. clarki 2. Neck black with pale rings, or with black recognized as the subfamily DipsadinaeÐfor bands several times wider than those pos- which the most diagnostic of several derived teriorly on body; eye moderate, going less hemipenial features is the distal division of than 2.5 times into loreal; total length less the sulcus spermaticus (Myers and Cadle, than 500 mm (Ͻ400 mm in two females) 1994: 27; Zaher, 1999: 33). The sulcus sper- ...... 3 maticus in the Dipsadinae usually divides ± Neck brown like body, with black transverse close to or within the capitulum; the hemi- lines not greatly wider than on rest of body penis of A. clarki is noncapitate and the sul- (®g. 1C); eye very small (®g. 3E), con- cus divides slightly below the midpoint of tained about three times in loreal plate; size the organ (at least in the retracted condi- large (one female 750 mm total length) . . tion)Ðsomewhat lower than in most dipsa- ...... A. depressiocellus dines but still higher than the basal bifurca- 3. Snout bluntly pointed in pro®le (®g. 9A, C); tion in the great majority of Xenodontinae (in loreal well separated from internasal; labi- which the sulcate bifurcation may rarely ap- als mostly white ...... 4 ± proach the midpoint of the organ, e.g., see Snout in pro®le rounded like a coral snake (®g. 9B); anterior corner of loreal ap- Zaher, 1999: ®g. 64, upper). proaching or touching corner of internasal; Atractus and Geophis have been often labials mostly black; neck black with sev- compared, but the closeness of their relation- eral pale rings, changing posteriorly to light ship is questionable and they are probably reddish brown ground color (red in life?) convergent to a similar life-style. On the basis of immunological comparisons, Cadle 7 A peculiarity shared by four of the ®ve Panamanian (1984) saw a closer relationship for Geophis specimens, and some others I have seen, is that the with Ninia and other Middle American dip- asymmetrical suture between the prefrontal plates is noticeably dextral to the internasal suture (®g. 3). The dex- sadines than with Peruvian Atractus elaps tral orientation also is seen in seven of eight species and A. major. Zaher (1999: 33) suggested, ®gured by Hoogmoed (1980). Some Atractus have the on the basis of a muscle character, an espe- two sutures aligned in the normal colubrid manner. 10 AMERICAN MUSEUM NOVITATES NO. 3391

Fig. 1. Panamanian Atractus: Holotypes in dorsal view. A. A. clarki Dunn and Bailey (MCZ 28800), ϫ1.0. B. A. darienensis, new species (KU 110274), ϫ0.8. C. A. depressiocellus, new species (AMNH 119876), ϫ0.4. D. A. hostilitractus, new species (AMNH 130330), ϫ0.8. See ®gure 10 for a ®fth Panamanian species (Atractus imperfectus, new species).

with black bars (®g. 1D) ...... Atractus clarki Dunn and Bailey ...... A. hostilitractus Figures 1A, 2A, 3A, 4, 5; map 1 4. Neck with black saddles (®g. 1B [obscure in life]) several times wider than posterior Atractus clarki Dunn and Bailey, 1939: 8±9 (ho- black bars, but narrow interspaces between lotype: MCZ 28800, an adult female collected anterior saddles brown like rest of ground by native worker in 1938, at Mine at Santa Cruz color, not whitish (or red or yellow); rostral de Cana [approx. 7Њ46ЈN, 77Њ41ЈW, 500 m], narrowly visible in dorsal view; eye length Province of DarieÂn, eastern Panama). less than 90% of distance to lip ...... A. darienensis DIAGNOSIS: Atractus clarki is readily dis- ± Neck black with pale bars or ringlike mark- tinguished from other Panamanian species in ings (®g. 10 [posterior pattern unknown]); rostral plate well visible in dorsal view; eye having a uniformly pale venter, narrow pale length about equal or slightly larger than dorsal bars, pale dashes on the lower scale distance to lip ...... rows, and in having a relatively large eye ...... Atractus imperfectus whose length is noticeably greater than its 2003 MYERS: SNAKES FROM PANAMA 11

Fig. 2. Panamanian Atractus: Holotypes in ventral view. A. A. clarki. B. A. darienensis. C. A. depressiocellus. D. A. hostilitractus. Specimens shown ϫ0.4±1.0 as given individually in ®gure 1. See ®gure 10 for the ®fth Panamanian species (Atractus imperfectus). distance to lip and which is contained less PROPORTIONS AND SCUTELLATION: Total than two times in length of the loreal plate. length 310 mm, tail length 36 mm (11.6% of Dunn and Bailey compared the species total). Slender, body rounded ventrolaterally; with Colombian A. pamplonensis and Ecua- body anteriorly slightly wider than high, but dorian-Peruvian A. collaris. Their nonillus- scarcely wider than head; body posteriorly trated original description is inadequate, par- becomes laterally compressed and higher ticularly in omitting mention of pale bars in than wide (about 6 mm high ϫ 5 mm wide the dorsal pattern and in not clearly stating near end of body); greatest head width 64.3% the presence of a well-de®ned nuchal collar. of head length from snout to end of parietals, about 2.0% of SVL; greatest body width REDESCRIPTION OF HOLOTYPE about 2.0% of SVL. Dorsal scales smooth, The specimen is a female, shown to be lacking apical pits, in 17±17±17 rows. Ven- adult by condition of the oviducts, which are trals about 181 (the estimated ``185'' in orig- large and convoluted. The neck was nearly inal description would have included three severed from the body by the collector, but large gulars or preventrals), anal plate undi- otherwise it is in good condition. vided, subcaudals in 33 pairs. 12 AMERICAN MUSEUM NOVITATES NO. 3391

Fig. 3. Heads of Panamanian Atractus; holotypes shown ϫ5, except ϫ2.7 for one large species. A. A. clarki Dunn and Bailey (MCZ 28800). B. A. darienensis, new species (KU 110274). C. A. hostilitractus, new species (AMNH 130330). D. A. imperfectus, new species (MCZ 50213). E. A. depressiocellus, new species (AMNH 119876). 2003 MYERS: SNAKES FROM PANAMA 13

TABLE 1 Standard Characters and Measurements (in mm) of Holotypes of Five Species of Panamanian Atractus

Head barely distinct from neck; snout greatest length 82% of loreal length; loreal bluntly rounded in dorsal view, bluntly point- long, 2.4 times longer than greatest height, ed in pro®le; rostral wider than high, well vis- well separated from internasal, entering eye; ible from above; internasals small, slightly no preoculars; supralabials 7, third and fourth wider than long, about half (53%) of length touching eye, fourth wider than high; two of suture between prefrontals; prefrontals postoculars, the uppermost largest; temporals large and nearly as wide as long (greatest pre- 1 ϩ 2, the upper one in row 2 elongated, frontal width 90% of greatest length); pre- reaching slightly past ends of parietals. frontal suture 63% length of frontal plate; su- Infralabials 7, ®rst pair in contact behind praoculars more than twice as long as broad; mental, ®rst three on each side in contact frontal as long as broad, roughly pentagonal with genials; single pair of large genials, in shape; interparietal suture 81% of frontal about three times longer than wide; three length. large median gulars or preventrals between Eye moderate, contained 1.4 times in lo- genials and ®rst ventral. Head plate tubercles real length, 2.7 times into snout length; eye tiny, sparse and inconspicuous. length 25% larger than its distance to lip; eye COLOR PATTERN: Dull blackish brown protuberant beyond edge of lip so that it is above, with about 25 narrow (about one scale slightly visible in ventral view. Nasal divided wide) whitish bars, extending from the ver- above and below near rear edge of naris, its tebral region down to the third or fourth scale 14 AMERICAN MUSEUM NOVITATES NO. 3391 row; these pale dorsal bars are mostly alternating on each side, but there are a few scat- tered pairs that medially connect and give appearance of a pale ring in dorsal view (e.g., ®rst pair on neck, ®g. 1A). The pale bars are vague but tend to be bordered by almost uniformly dark areas about 1±2 scales wide, which add a little emphasis. Most intervening dorsal scales have pale centers, as do all lateral scales in rows 1±3, resulting in a nearly continuous white line on the second row. The whitish line continues well onto the tail, which dorsally has vague indications of the pale body bars. The snout is gray, from its tip to the middle of the prefrontals above and to the eyes laterally. The top of the head is darker brown back to the middle of the parietals, with this color extending ventrad to include the postoculars and circling narrowly under the eyes. The supralabials are mostly white. An ill-de- ®ned black bar extends obliquely from the front edge of the primary temporal onto supralabial 6, with a small isolated spot match- ing up with it below the mouth, on the anterior end of the last infralabial. Dorsally an orange-brown band crosses the rear of the head, being narrowed medially and slightly broken by brown pigment along the interparietal suture; laterally this pale orangish Fig. 4. The second known specimen of Atrac- band is con¯uent with, but distinct from, the tus clarki (MCZ 13301(), from an unspeci®ed white areas of the primary temporal and pos- locality in the Colombian ChocoÂ, ϫ1.2. terior labials. Except for the small vague dark spot on the last infralabial, the under- process. An expanded ¯ange on maxilla ex- side of the head is immaculate pale yellow. tending mediad and ventrad adjacent to the The ventrals and subcaudals are slightly two small posterior teeth. Ectopterygoid edged with the dark body color and there is forked, with one branch (apparently an ex- a weak, irregular median line of brown pig- panded ¯ange) tightly bound against the ex- ment under the tail. The venter otherwise is panded maxillary ¯ange. Maxillary process immaculate pale yellow. of palatine absent. MAXILLO-PALATO-PTERYGOID ARCH: Ex- amined in situ on right side. Maxilla arched, ASECOND SPECIMEN OF ATRACTUS CLARKI extending anteriorly nearly to suture between rostral and ®rst supralabial, with 6 well- A Colombian specimen (MCZ 13301) spaced recurved teeth, decreasing in size from the ChocoÂ, without further locality data, posteriorly. First tooth springs from anterior has for many decades been misidenti®ed as tip of maxilla; ®rst four teeth large, with no- Atractus multicinctus, which is a better ticeable gaps (not sockets) between 2 and 4, known Chocoan species. This ``new'' speci- followed by a larger gap and two small teeth, men (®g. 4) is a male, judged adult because the ultimate noticeably smaller than penulti- the hemipenial spines are calci®ed. It was mate; teeth distally angular in cross section, collected by M.V. Campbell in 1919. with a lateral edge. Maxilla extending pos- PROPORTIONS AND SCUTELLATION: MCZ teriorly past small teeth as a short toothless 13301 is 278 mm in total length, 43 mm tail 2003 MYERS: SNAKES FROM PANAMA 15 length (15.5% of total), with 159 ventrals and 40 pairs of subcaudals. Compared with the female holotype it has 22 fewer ventrals, 7 more caudals, and a relatively longer tail, with these differences being in the direction of expected sexual dimorphism. Other scale counts are identical to the holotype, and head plate proportions are similar. This specimen has a relatively wider head (71.6% of length vs. 64.3% in holotype), with the snout ap- pearing more pointed in dorsal view. The eye is contained 1.25 times (1.4 times in holotype) in the loreal, which is 2.1 (vs. 2.4) times longer than greatest height, and contained 2.8 (2.7) times in the snout length. Eye length is 25% larger than distance to lip (identical to holotype), but the eyes are not quite visible from below as they are in the holotype, possibly due to the relatively wider head. As in the female holotype, the body anteriorly is slightly wider than high but posteriorly becomes laterally compressed and higher than wide. COLOR PATTERN: The color pattern of the Colombian specimen (®g. 4) is bolder than on the Panamanian holotype (®g. 1A) but is essentially the same: There are about 28 ver- tical white bars on the left side, mostly alternating but sometimes aligned with those on the right; none of these are medially fused and most do not reach the vertebral scale row; there is an aberrant longitudinal patch of white on the right side of neck in place of Fig. 5. Hemipenis of Atractus clarki (MCZ the usual bars. As on the holotype the lateral 13301(). Uneverted left organ opened midven- white bars are vaguely set in areas of uni- trally, ϫ10. formly pigmented scales, which otherwise contain pale dashes. The lower several scale rows are strongly dashed with white. two slips of retractor muscle merging at the The pale nuchal collar includes the dorsal end of subcaudal 11. The organ was cut open scales immediately behind the parietals and along its midventral line, removed, and upper secondary temporals, and the collar is pinned ¯at for illustration (®g. 5). thus wider than on the holotype. The snout The distal third of the organ is bilobed and seems to have been gray as in the type. The calyculate, with the well-de®ned calyces ill-de®ned black postocular bar described for bearing blunt to slightly pointed soft papillae. the holotype is lacking, being represented on The calyculate region is continuous from this specimen only by a few isolated small lobe to lobe at the medial side of the crotch; pigment spots. There is a smudge of dark pig- the area within the fork of the sulcus is spi- ment on the tip of the chin and a rare dark nose. The sulcus spermaticus divides less speck on ventrals and subcaudals, but other- than halfway (about 45%) up the organ, and wise the ventral surfaces are uniformly pale. its branches extend to the tips of the lobes in HEMIPENIS: The left retracted hemipenis bi- a centrifugal orientationÐone branch lying furcated at the level of subcaudal 8 and ter- on the ventral wall of the ventral lobe (this minated at the end of subcaudal 10, with the branch cut across when lobe opened midven- 16 AMERICAN MUSEUM NOVITATES NO. 3391 trally, ®g. 5) and the other lying on the dorsal tus are ``bilobed at tips'' and ``never calycu- wall of the dorsal lobe (determined by medial late.'' However, the organ of A. clarki is bi- incision of this lobe, concealed in the ®gure); lobed for a third its length and the lobes are I am uncertain whether the centrifugal ori- markedly calyculate. Hoogmoed (1980) ex- entation of the branches would be maintained plicitly mentioned the existence of calyces in on the everted organ. The midsection of the several species of Atractus in Surinam, al- hemipenis is densely spinose, with a few though in one instance (p. 23) he described dozen medium-sized spines; most of these them as ``scalloped calyces (Savage, 1960, spines are arranged in V-shaped rows broken ®g. 4A)''Ðthis reference being to structures by the sulcus spermaticus, but smaller spines described by Savage (1960: 24) not as calyces lateral and basal to the aforesaid are nearly but as ``scalloped transverse ¯ounces'' (which arranged in transverse rows. There is no ev- conceivably might be derived from calyces by idence of capitation or semicapitation be- loss of the vertically aligned tissue ridges). tween the spinose midsection and calyculate Fernandes (``1995'' [1996]: 43, 50, ®g. 1) lobes. The extreme base of the organ is nude, added the trait of hemipenial capitation with with longitudinal expansion folds. Above the calyculation for the Brazilian A. reticulatus nude part are a few spinules and, most con- and some other southern Atractus, and Fer- spicuously, a cluster of small spines atop a nandes et al. (2000: 3, 5) described the Bra- heavy longitudinal fold on the dorsal wall. zilian A. maculatus and A. zebrinus as having There is a heavy, nearly nude transverse fold hemipenes ``capitate, bilobate . . . the distal of tissue adjacent to the spinose cluster; this portion covered by spinulate calyces''. fold proximally overlays a small cavity con- The taxonomic distribution and nature of taining a few concealed spines (indicating both capitation and calyces/¯ounces within that the cavity is everted when the organ is), Atractus need to be revisited (male genitalia and distally the transverse fold marks the have been comparatively examined for less base of a deep, lateral naked pocket. (Thus, than half of the currently recognized species the longitudinal lateral pocket does not con- [mainly by Savage, 1960, and Hoogmoed, tinue to the base of the hemipenis as it does 1980] and illustrations are regrettably rare). in most snakes having such a structure, but Calyces are probably symplesiomorphic for it almost certainly retains its identity after both xenodontine and dipsadine colubrids, eversion of the hemipenis.) with absence being a presumed secondary loss, as in the tribe Xenodontini (Myers, REMARKS 1986: 6) and in many or most Atractus. When Savage's seminal study is extended to Dunn and Bailey (1939) named this spe- cover the entire genus, it may be found that cies after Herbert C. Clark, ®rst director of Atractus clarki is simply primitive in the re- the Gorgas Memorial Laboratory and insti- tention of relatively long, hemipenial lobes gator of the Panamanian snake census. Santa bearing typical calyces. See page 9 for a Cruz de Cana, the type locality of Atractus comment on the bifurcation of the sulcus clarki, is situated at 500 m elevation (my al- spermaticus. timeter reading) on the eastern ¯ank of the Atractus clarki also seems primitive in its SerranõÂa de PirreÐroughly at 7Њ46ЈN, relatively large eye, which is suf®ciently pro- 77Њ41ЈW (map 1). It is the site of gold min- tuberant so as to be visible (holotype) or al- ing that has been carried on periodically most visible (MCZ 13301) in ventral view, since the end of the 16th century, being usu- more like a generalized terrestrial colubrid ally deserted and reverting to jungle in the than the majority of Atractus, most of which intervening periods (Myers, 1969: 22). seem to be semifossorial on morphological The second specimen and ®rst male of grounds. Atractus clarki, from somewhere in the Col- ombian ChocoÂ, corroborates the diagnostic Atractus darienensis, new species features of the species but might be perceived Figures 1B, 2B, 3B, 6, 9A; Map 1 to call into question its generic status. Savage HOLOTYPE: KU 110274 (®eld no. CWM (1960: 29) stated that the hemipenes of Atrac- 6095), an adult female caught by C.W. Myers 2003 MYERS: SNAKES FROM PANAMA 17

Fig. 6. Atractus darienensis, new species. The holotype in life (KU 110274). The re¯ective scale surfaces cause the color pattern to be less evident in life than in preservative (compare with ®g. 1B). on January 25, 1966, at 500 m above sea fectus in having white labials, a similarly level on the north end of the SerranõÂa de Pir- vague nuchal collar, and a similar snout re, Province of DarieÂn, eastern Panama. The shape in pro®le (compare ®g. 9A and 9C). type locality, a temporary forest camp, is sit- Atractus imperfectus, however, has pale bars uated roughly at 8Њ00ЈN, 77Њ43ЈW (map 1; or broken rings on a black neck (®g. 10 [pos- ®gs. 7, 8). terior pattern unknown]), dark serration ETYMOLOGY: The speci®c name is an ad- across upper edges of the supralabials, a rel- jective derived from the eastern Panamanian atively larger eye, broader snout, and a dor- Province of DarieÂn ϩ the adjectival suf®x sally more extensive rostral plate (®g. 3). -ensis (indicating place of origin). These differences are elaborated under DIAGNOSIS: Atractus darienensis can be Atractus imperfectus. distinguished from other Panamanian and most South American species by the pattern DESCRIPTION OF HOLOTYPE of black markings on a medium brown (reddish brown in life) ground color. There are a The holotype is a sexually mature female few black saddles on the neck which are sev- as shown by enlarged, ¯abby (nonconvolut- eral times wider than the alternating or me- ed) oviducts that appear recently spent. dially connected black bars that occupy the PROPORTIONS AND SCUTELLATION: Total rest of the body; the shiny scales make this length 346 mm, tail length 33 mm (9.5% of pattern less obvious in life (®g. 6) than when total). Moderately proportioned, with body the snake is immersed in liquid (®g. 1B). noticeably wider than head, wider than high Atractus hostilitractus also has black mark- and rounded ventrolaterally; greatest head ings that are anteriorly wider, but these are width 78.9% of head length from snout to partially separated by pale rings, and the pos- end of parietals, about 1.8% of SVL; greatest terior bars are less numerous than in dari- body width about 2.8% of SVL. Dorsal enensis and set on a much lighter ground col- scales smooth, lacking apical pits, in 17±17± or (®g. 1D); A. hostilitractus also has black 17 rows. Ventrals 159, anal plate undivided, supralabials, and the snout in pro®le is re- subcaudals in 25 pairs. mindful of a coral snake (®g. 9B). Head barely wider than neck; snout blunt- Atractus darienensis resembles A. imper- ly rounded in dorsal view, bluntly pointed in 18 AMERICAN MUSEUM NOVITATES NO. 3391

Fig. 7. Northern end of the SerranõÂa de Pirre, as seen looking SSE from a distance of 15±20 km at El Real, 20 m elevation (December 1965). The twin breastlike peaks were determined by altimeter to be 960 m and 1000 m (from left to right); the still higher prominence (1110 m) farther south is Cerro Cituro (wrongly identi®ed on some maps as ``Cerro Pirre'', which lies still farther south at about 1550 m). Arrow indicates general region of the type locality of Atractus darienensisÐa temporary base camp at 500 m elevation, established prior to exploring higher reaches of the serranõÂa. See ®gure 8. pro®le; rostral wider than high, scarcely vis- below at its posterior edge, its greatest length ible from above; internasals small, as wide 54% of loreal length; loreal long, 3.3 times as long, less than half (43%) of length of longer than greatest height, well separated prefrontal suture; prefrontals large, wider from internasal, entering eye; no preoculars; than long (greatest prefrontal width 130% of supralabials 7, third and fourth in eye, fourth greatest length); prefrontal suture 74% length somewhat wider than high; two subequal of frontal plate; supraoculars large, nearly as postoculars; temporals 1 ϩ 2, the upper one broad as long; frontal as long as broad, in row 2 elongated, extending past end of roughly pentagonal in shape; estimated in- interparietal suture. terparietal suture equal to prefrontal suture, Infralabials 7, ®rst pair in contact behind 74% of frontal length. The parietals are ab- mental, ®rst four on each side in contact with errantly fused with each other and with the genials; single pair of large genials, 2.4 times right posterior half of the frontal plate (®g. longer than wide; four large medial gulars 3B). and preventrals between genials and ®rst Eye small, contained 2.2 times in loreal ventral. Head plate tubercles tiny, sparse, and length, 3.3 times in snout length; eye length inconspicuous. smaller (86%) than distance to lip; eyes set COLOR PATTERN: In life the holotype of close to head, not protruding to edge of lips, Atractus darienensis was dark reddish brown thus concealed from ventral view. Nasal above with black dorsal saddles and bars, weakly divided above middle of naris and which, as indicated in my ®eldnotes, were 2003 MYERS: SNAKES FROM PANAMA 19

Fig. 8. Type locality of Atractus darienensis, at 500-m base camp in forest on northern end of the SerranõÂa de Pirre (see ®g. 7). Early dry season (January 14, 1966); within two weeks after this photo- graph, the saturated ground surface dried and there was abundant leaf fall, with the forest seeming to be nearly 50% deciduous. The crest of the serranõÂa supports dense cloud forest (illustrated in Myers, 1982). not very evident because the smooth shiny terspaces are only 1±2 scales wide between scales readily re¯ected light, so as to reduce the ®rst few saddles, but posteriorly the in- contrast in the dorsal pattern (compare life terspaces become somewhat wider than the view in ®g. 6 with ®gs. 1B, 2B). Labials and cross-bars. Very irregularly shaped grayish anterior venter pale yellow, turning still paler black lateral blotches on rows 1±3, alternat- yellow on posterior half of venter, with ill- ing with the dorsal cross-bars, appear poste- de®ned black ventral markings and with anal rior to the neck saddles, with each lateral plate and underside of tail black. Tiny iris blotch occupying an area equal to several red. scales. The last several lateral blotches are In preservative (®gs. 1B, 2B) medium longitudinally fused, forming a poorly de- brown with about 33 black dorsal markings: ®ned black stripe anterior to the tail. The The ®rst six are complete saddles (touching short tail has half a dozen irregular black scale row l on each side), whereas the pos- bars. terior markings are shorter (to rows 3 or 4) The head is black above, with this color and several are asymmetrical, forming alter- being medially continuous with the ®rst neck nating dorsolateral bars on each side of the saddle and irregularly edging the upper edges vertebral line. The ®rst three black saddles of the supralabials, which are mostly white. are 6±8 scales wide, the next three are 3±4 An oblique extension of the black head cap scales wide, and posteriorly the dorsal cross- extends posteroventrally across the rear of bars are about 2 scales wide; the brown in- supralabial 6 and most of 7 and barely onto 20 AMERICAN MUSEUM NOVITATES NO. 3391 the ultimate infralabial. Posterior to this an tober 31, 1974, Cerro Azul [Cerro Jefe] re- irregular anterodorsal extension of the pale gion, Province of PanamaÂ, Panama. The type throat color forms an incomplete nuchal col- locality is in the general region of Cerro Jefe larÐwhite ventrally, pale brown dorsallyÐ on the Piedras-Pacora Ridge, at about ending on the posterolateral edge of each pa- 9Њ14ЈN, 79Њ23ЈW (see Remarks). rietal. ETYMOLOGY: This snake has conspicuously The underside of the head and throat are depressed, very small eyes, and hence the white with sparse brown markings; the ven- speci®c name, a noun in apposition, is deter (®g. 2B) is heavily clouded with black rived from the Latin adjective depressus (low and brown pigmentation; the anal plate and or depressed) ϩ the noun ocellus (a little underside of the tail are black. eye). MAXILLO-PALATO-PTERYGOID ARCH: Ex- DIAGNOSIS: Atractus depressiocellus is im- amined in situ on right side. Maxilla arched, mediately distinguished from other Pana- extending anteriorly to middle of ®rst su- manian Atractus by its large size, relatively pralabial, with 7 well-spaced recurved teeth, uniform dorsal pattern of irregularly paired decreasing in size posteriorly (a large re- black cross-lines on a brown ground, and placement tooth close behind and slightly very small eyes that are notably depressed medial to the fourth tooth is not included in on a broad, relatively short head. It is the this count). First tooth springs from anterior only Panamanian Atractus in which head tip of maxilla; ®rst ®ve teeth large, with no- length (tip of snout to end of parietals) is ticeable gaps (not sockets) between 2 and 5, equalled by head width (®g. 1C). followed by a larger gap and two very small In total length, Atractus depressiocellus teeth (ultimate smallest); teeth distally an- approaches A. obesus from Andean Colom- gular in cross section, with an anterolateral bia and the Amazonian A. major and A. tor- edge. Maxilla extending posteriorly past quatus. These differ from depressiocellus in small teeth as an elongate toothless process. color pattern (photographs in Marx, 1960; An expanded ¯ange on maxilla extending Martins and Oliveira, 1993) and in various mediad and ventrad adjacent to the two small features of scutellation. Atractus depressi- posterior teeth. Ectopterygoid forked, with ocellus also differs noticeably from these in one branch forming an apparently expanded its very small depressed eye and in the as- ¯ange that is tightly bound against the ex- sociated feature of exceptionally tall suprala- panded maxillary ¯ange. Maxillary process bials 3±5 (®g. 3E). It may be unique in either of palatine absent. lacking or in having exceptionally tiny internasals (see below). REMARKS I found the holotype of Atractus darienen- DESCRIPTION OF HOLOTYPE sis at night in dense forest (®g. 8), as it was As suggested by its large size and proved crawling across a small bare spot on the for- by large ovaries and large convoluted ovi- est ¯oor. The species is quite distinct from ducts, the specimen is a sexually mature fe- its geographically nearest neighbor, A. clarki, male. It is rather soft, with internal organs whose type locality is only about 25 km to poorly preserved; color pattern and other ex- the south. Both type localities lie at 500 m ternal features are clearly evident except that elevation on the lower slopes of the SerranõÂa the rostral region was rubbed raw or injured de Pirre (map 1), in an evergreen seasonal in life. The rostral plate and anterior parts of forest (monsoon rain forest) having a Janu- the nasal plates consequently are lacking, and ary±March dry season during which many presence of internasals cannot be determined, trees are deciduous (Myers, 1969, 1982). although the internasals would be exceptionally small if present (®g. 3E). Atractus depressiocellus, new species PROPORTIONS AND SCUTELLATION: Total Figures 1C, 2C, 3E; map 1 length 750 mm, tail length 77 mm (10.3% of HOLOTYPE: AMNH 119876, an adult fe- total). A relatively long and heavy-bodied male collected by George Barratt, Jr., on Oc- snake, but body not any wider than the broad 2003 MYERS: SNAKES FROM PANAMA 21 head; body wider than high and rounded ven- als and ®rst ventral. Head plate tubercles trolaterally; greatest head width slightly larg- tiny, sparse, and inconspicuous. er (104%) than head length from snout to end COLOR PATTERN: Head and body light yel- of parietals, about 2.9% of SVL; greatest lowish brown, turning slightly grayish brown body width about 2.8% of SVL. Dorsal on rear of body and tail, with many irregular scales smooth, lacking apical pits, in 17±17± black dorsal cross-bars (®g. 1C). Despite 17 rows. Ventrals 167, anal plate undivided, some irregular interconnections, the black subcaudals in 30 pairs. bars tend to be paired, with the lateral ends Head broad, wider than neck; snout broad of each pair converging and often connecting and bluntly rounded in dorsal view, rounded on about the third and fourth scale rows. The in pro®le; missing (damaged) rostral proba- are about 38 ``pairs'' of such markings on bly was somewhat wider than high; missing the body, with the last three individually internasals either were very small if origi- fused to form single black markings anterior nally present (see above), or possibly fused to the tail, which has small irregular black completely with large prefrontal plates; pre- markings. The dorsal areas contained within frontals very large, longer than wide (great- a pair of markings are somewhat shorter than est prefrontal width 67% of greatest length); the interspaces between pairs but are undif- prefrontal suture longer (118%) than length ferentiated from the overall yellowish brown of frontal plate; supraocular large, a little ground color. There are small ventrolateral wider than long; frontal short, noticeably black spots between and below some of the wider than long (length 76% of width), paired dorsal bars but they are irregular and roughly pentagonal in shape; interparietal su- show no de®nite pattern. ture much longer (122%) than length of fron- The dorsal black bar on the nape is chev- tal plate but only slightly longer than pre- ron-shaped and anteriorly connected with a frontal suture. vague and irregular black blotch between the eyes. A black spot on top the left side of the Eye very small, contained 3.3 times in lo- head lies on the anterior half of the elongated real length, 4.4 times into snout length; eye secondary temporal and adjacent part of the length less than half (46%) its distance to lip; parietal, but there is only a faint outline of eye set in a pronounced depression, with su- this spot on the right side. Lower postocular praocular dropping almost vertically to eye on left side and both postoculars on right are and with all other circumocular plates slant- black, and just in front of the eye there is a ing inward to eye. Nasal weakly divided suffusion of black on the posterior ends of above and below posterior part of nasal, an- the loreal and third supralabial. Otherwise teriorly damaged but with a greatest length the upper parts of the head are colored like roughly 28% of loreal length; loreal long, 2.7 the body; supralabials are pale brownish. times longer than greatest height, entering Underside of head and throat pale brown- eye; no preoculars; supralabials 7, third and ish, sparsely and irregularly marked with fourth in eye; correlated apparently with re- black. Ends of the ventral plates are light duced eye size, supralabials 3±5 are very tall brown like the dorsum, but the venter (®g. (basal widths of labials 4 and 5 are about 2C) is mostly dark owing to a wide black 72% and 60% of greatest height); two sub- marking that occupies the major part of ev- equal postoculars; temporals 1 ϩ 2, the upper ery ventral (including anal plate) except in one in row two elongated, extending slightly the neck region, where these markings are past ends of parietals, and (on each side of smaller and less regular. The subcaudals are head) with a concavity in posterior edge sug- black except for lateral encroachments of gesting fusion with one or more dorsal light brown. scales. MAXILLO-PALATO-PTERYGOID ARCH: Ex- Infralabials 7, ®rst pair in contact behind amined in situ on right side. Maxilla robust, mental, ®rst three on each side in contact not very arched, extending anteriorly past su- with genial; single pair of large genials, 2.2 ture between ®rst two supralabials, with 7 times longer than wide; four large preven- recurved teeth. First tooth springs from an- trals (gulars wider than long) between geni- terior tip of maxilla; ®rst ®ve teeth subequal, 22 AMERICAN MUSEUM NOVITATES NO. 3391 relatively large, with small gaps between 2 dinary that it has escaped detection for so and 5, followed by a larger gap and two long. sockets for two small teeth (size con®rmed on left side); teeth distally angular in cross Atractus hostilitractus, new species section, with a lateral edge. No posterior Figures 1D, 2D, 3C, 9B; map 1 toothless process of maxilla. An expanded ¯ange on maxilla extending mediad and ven- HOLOTYPE: AMNH 130330, an adult fe- trad adjacent to the two small posterior teeth. male collected by TomaÂs Quintero on Octo- Ectopterygoid probably forked, with one ber 29, 1967, at ``Morti Hydro'' [about 100± branch forming an expanded ¯ange tightly 200 m elev., at 8Њ52Ј28ЉN, 77Њ54Ј19ЉW], RõÂo bound against the expanded maxillary ¯ange. MortõÂ, Province of DarieÂn, Panama. Maxilla tightly bound to palatine by liga- The type locality was a temporary camp mentous connections, but maxillary process and helicopter pad operated by the Interoce- of palatine seemingly absent. anic Canal Survey from October 1966 to about 1969. Coordinates and elevation were determined from Duke and Porter (1970: 10) REMARKS in conjunction with a 1964 topographic map The collector was a former ®eld assistant (1:50,000 Mulatupo, sheet 1388-IV). The lo- of the Gorgas Memorial Laboratory, which cality is 2.7 km NNE of the Cuna village of is relevant to placing the type locality of MortõÂ Arriba, and 7.8 km SW of Camp Sum- Atractus depressiocellus. ``Cerro Azul'' of mit (DemartaganÄala), another former helipad GML ®eld parties was essentially equivalent and collecting site on the low continental di- to Cerro Jefe, the 980-m high point on the vide (Myers, 1969: ®g. 13). Piedras-Pacora Ridge about 37 km NE of ETYMOLOGY: The speci®c name, a noun in downtown Panama City (Myers, l969: 27). I apposition, is compounded from the Latin have added the word ``region'' to the type hostilis (hostile) ϩ the noun tractus (region locality because of a belief that specimens or territory). The associative implication is to labeled ``Cerro Azul'' might have been taken the 19th- and early 20th-century history of a anywhere from about 5 km southwestward of part of the Panamanian interior that includes Cerro Jefe to about 5 km to the north. The the RõÂo MortõÂ.8 likely elevational range for most reptiles col- DIAGNOSIS: Atractus hostilitractus is char- lected is about 200±800 m in my estimation. acterized by a distinctive, complex color pat- Atractus depressiocellus is morphological- tern (®gs. 1D, 2D), the irregularity of which ly quite divergent from the other Panamanian gives the appearance of possibly being high- species, particularly in the depressed and ly variable in life. The holotype of A. dari- very small eye and in the internasals, which enensis has similar numbers of ventrals and are either lacking or much reduced in size. subcaudals and seems arguably closest to The dentition is also different from the other four Panamanian species, which have the 8 The type specimen of Atractus hostilitractus was maxilla strongly arched and bearing well- collected by my former assistant TomaÂs Quintero in the spaced, ®rmly set teeth that decrease in size ``forbidden land'', where, as mentioned by Terry (1956: posteriorly. In contrast, the maxilla of A. de- 19, 66), ®eldwork had been long discouraged. Through much of the ®rst half of the 20th century, the inland pressiocellus is not strongly arched and the Cuna Indians rather successfully excluded most outsid- anterior teeth are not as widely spaced and ers from travelling past the mouth of RõÂo MortõÂ on the are subequal in size; also, the maxillary teeth upper RõÂo Chucunaque. Caution and diplomacy were of depressiocellus are not as ®rmly anky- needed even as recently as late 1965, when Quintero and I worked along the upper Chucunaque above the village losed as in the other species and are easily of MortõÂ Abajo. This atmosphere changed considerably dislodged from their sockets (due partly to starting in 1966, when the politically independent but the soft state of preservation of the holo- numerically small group of inland Cuna (Torres de Ar- type?). auÂz, 1970: 46±63, 169) had to acquiesce to an in¯ux of outsiders during survey of ``Route 17'' of the sea-level With a total length of 750 mm (673 mm canal feasibility studies. This route, explored for possi- SVL), Atractus depressiocellus ranks among ble excavation by nuclear explosives, included the RõÂo the largest of Atractus, and it seems extraor- MortõÂ (Commission, 1971). 2003 MYERS: SNAKES FROM PANAMA 23

The color pattern of A. hostilitractus also is suggestive of the Andean A. sanguineus and A. wagleri (and perhaps A. andinus as well), as based on poorly reproduced photographs and brief type descriptions (Prado, 1944, 1945, also 1946).9 But compared with hostilitractus, those species have relatively longer tails (12.9±13.7% of total length), many more black bars, and little differentia- tion between neck and body patterns.

DESCRIPTION OF HOLOTYPE It is a sexually mature female, with ova Ͼ4 mm and enlarged convoluted oviducts. Part of the posterior body was damaged during capture, but the specimen is well preserved. PROPORTIONS AND SCUTELLATION: Total length 374 mm, tail length 40 mm (10.7% of total). Moderately proportioned, with body rounded ventrolaterally and somewhat wider than head; body wider than high anteriorly but laterally compressed and higher than wide posteriorly (probably not an artifact of preservation); greatest head width 77.7% of head length from snout to end of parietals, 2.3% of SVL; greatest body width about 2.5% of SVL. Dorsal scales smooth, lacking apical pits, in 17±17±17 rows. Ventrals about 160 (including 10 estimated missing in 17 Fig. 9. Heads of three new species of Pana- mm of damaged body), followed by a half- manian Atractus, all holotypes ϫ3.8. A. A. dari- ventral in front of undivided anal plate; sub- enensis. B. A. hostilitractus. C. A. imperfectus. caudals in 27 pairs. Head slightly wider than neck; snout short hostilitractus in general color pattern (®g. and broad, bluntly rounded in dorsal view, 1B), but pronounced differences in snout rounded in pro®le (®g. 9B); rostral wider shape (compare ®g. 9A and 9B) are strong than high, well visible in dorsal view; inter- evidence that these snakes are not color morphs of a single species. The close prox- 9 Based on the descriptions and illustrations, one sus- imity of the loreal and internasal (in narrow pects that A. sanguineus and A. wagleri, and perhaps A. andinus (see especially Prado, 1946), might be closely contact in the holotype) may also be a dis- related geographic replacements, if not actually conspe- tinctive feature of A. hostilitractus (®g. 3C). ci®c. The last possibility seems less likely, since they The wide-spaced bars in the posterior dor- come from different cordilleras. The ®rst is from Yaru- sal pattern of A. hostilitractus, and the pres- mal (2300 m) in the northern end of the Cordillera Cen- ence of a black vertebral streak, are sugges- tral; the second is from Humbo (Ͻ1000 m), nearly 200 km to the southeast, on the western side of the Cordillera tive of the red and black A. schach of Suri- Oriental; the third is from Andes (1500 m), some 150 nam and Brazil, but that is a smaller snake km SSW of Yarumal in the Cordillera Occidental. (Ͻ300 mm total length) which does not have Dunn (1944: 398), incidentally, seems to have con- the neck conspicuously different from the sidered without explanation that the locality Humbo (5Њ36ЈN, 74Њ17ЈW) is synonymous with Muzo (5Њ32ЈN, body (save for one wider saddle) and which 74Њ06ЈW), a bit of misinformation perpetuated in Med- has less black pigmentation ventrally (Hoog- em's (1965: 341) gazetteer and in PeÂrez-Santos and Mo- moed, 1980: 31, color pl. 1d). reno (1988: 106). 24 AMERICAN MUSEUM NOVITATES NO. 3391 nasals small, as wide as long, nearly half 13 scales long on the vertebral line and pos- (48%) length of prefrontal suture; prefrontals teriorly bordered by a 1±2-scale-wide whitish large, longer than wide (greatest prefrontal (yellow in life?) ringlike interspace that is width 86% of greatest length); prefrontal su- medially broken by two black scales; at its ture 70% length of frontal plate; supraoculars midpoint the ®rst black band contains a pair large, slightly longer than broad; frontal of vague reddish spots, each occupying parts slightly wider than long, roughly triangular of several scales on rows 4±7. (2) The sec- in shape; interparietal suture slightly longer ond band is 12 scales long and posteriorly than prefrontal suture, 76% of frontal length. bordered by a medially broken, whitish ring- Eye small, contained 2.25 times in loreal like interspace whose halves are offset and length, 3 times in snout length; eye length separated by the black vertebral scales; the smaller (86%) than distance to lip; eyes set middle of this band contains a weak reddish close to head, not protruding to level of edge suffusion several scales wide between rows of lips, hence not visible in ventral view. Na- 2 and 3 on each side. (3) The third band is sal divided behind rear edge of naris, its 8±10 scales long and posteriorly bordered by greatest length 56% of loreal length; loreal an unbroken 1±2-scale-wide whitish ringlike long, 3.0 times longer than greatest height, interspace; there is a weak reddish suffusion, in narrow contact with posterolateral corner especially on the right side, in the middle of of internasal, entering eye; no preoculars; su- the band. (4) The fourth band is 7 scales long pralabials 7, third and fourth in eye, fourth with some reddish spotted scales in the mid- slightly wider than high; two postoculars, up- dle. per the largest; temporals 1 ϩ 2, the upper Posterior to the pale ``rings'' and four one in row 2 elongated, extending past ends broad black bands, the dorsum is light red- of parietals. dish brown with about 20 black bars on each Infralabials 7, ®rst pair in contact behind sideÐthe bars on opposite sides mostly al- mental, ®rst three (right) or four (left) in con- ternate and frequently interconnect at their tact with genials; single pair of large genials, corners, but in a few places they are aligned 2.2 times longer than wide; four large me- and medially connected to form narrow dian gulars and preventrals between genials cross-bands. The ®rst few bars are wider than and ®rst ventral. Head plate tubercles minute the others and the ®rst six extend from the but well distributed and easily seen under midline to scale row 1, whereas those pos- magni®cation. terior end at about row 2. Most bars are COLOR PATTERN: The dorsal pattern of about 2 scales wide and separated by wider Atractus hostilitractus is complex (®g. 1D). interspaces up to 6 scales wide. The scales The holotype at a glance has a black neck in the reddish brown ground color are mod- with pale rings, followed by irregular black erately peppered with melanophores, a heavi- bars widely spaced on a light reddish brown er concentration of which form a dark streak body.10 I interpret the anterior black color to on the vertebral row. However, individual represent several partially connected wide scales adjacent to the black bars are whitish cross-bands as follows: (1) The ®rst band, and have fewer melanophoresÐsuggestive anteriorly fused with the dorsal head color, is of the pale yellow edging on the black rings of some coral snakes. 10 The appearance of the reddish brown color leads The head is overall black, with vague me to suspect that it was red in life, and that the whitish whitish spots on the supralabials and infrala- ringlike dorsal markings on the neck may have been bials and with a wide, incomplete nuchal col- yellow, possibly also with weak yellow edging on the lar that extends up behind the mouth to end posterior black bars. The ground color resembles the faded red of many preserved coral snakes, in which yel- on the temporal plates (®g. 9B). The chin, low rings fade to white. If in life it is red and black, or venter, and underside of the tail are black red, yellow, and black, A. hostilitractus might arguably (®g. 2D), except for a ventrolateral white be called a ``false coral'', although the pattern match streak that extends posteriorly from the nu- with broadly sympatric models (e.g., Micrurus clarki) would be very crude compared with many coral snake chal collar to the middle of the second black mimics (e.g., see color ®gures in Greene and Mc- neck band, thence discontinuously to the re- Diarmid, 1981). gion of light reddish brown body color, 2003 MYERS: SNAKES FROM PANAMA 25

Fig. 10. Atractus imperfectus, new species. Dorsal and ventral view of the only known specimen (MCZ 50213, holotype), ϫ2.8. Collected for the old Panamanian snake census in 1936, this imperfect specimen was the ®rst Atractus from Panama. which edges the ventrals on most of the and neck [now MCZ 50213, the holotype] from body. the Piedras-Pacora Ridge, central Panama). MAXILLO-PALATO-PTERYGOID ARCH: Ex- Atractus crassicaudatus: Smith, 1958: 223 (coun- amined in situ on right side. Maxilla strongly try checklistÐthis usage followed by subse- arched, extending anteriorly to suture be- quent authors of regional lists). Atractus sp.: Savage, 1960: 32 (`ùndescribed tween rostral and ®rst supralabial, with 7 population allied to Atractus crassicaudatus''). well-spaced recurved teeth, decreasing in size posteriorly (®rst several teeth subequal). HOLOTYPE: MCZ 50213, head and neck First tooth springs from anterior tip of max- only (sex ?), killed by a surveying party in illa; ®rst ®ve teeth large, with noticeable gaps 1936 for the Panama snake census, from the (not sockets) between 2 and 5, followed by Piedras-Pacora Ridge, [Province of PanamaÂ], another gap and two small teeth (the ultimate Panama. (Precise locality unknown, but a one smallest); teeth distally angular in cross more-or-less central point on this divide is section, with a lateral edge. Maxilla extend- 9Њ16ЈN, 79Њ20ЈW.) ing posteriorly past small teeth as an elongate ETYMOLOGY: The speci®c name imperfec- toothless process. An expanded ¯ange on tus is a Latin adjective used in allusion to the maxilla extending mediad and ventrad adja- condition of the holotype. cent to the two small posterior teeth. Ecto- DIAGNOSIS: Although known only from a pterygoid forked, with one branch an evi- partial specimen (®g. 10), this species is dis- dently expanded ¯ange that is tightly bound tinguished from other Panamanian Atractus against the expanded maxillary ¯ange. Max- as follows: Resembles clarki in having pale illary process of palatine absent. bars on black neck but differs in larger size, relatively smaller eye, extensive black pig- Atractus imperfectus, new species mentation on anterior ventrals, and in having Figures 3D, 9C, 10; map 1 only the hint of a pale collar. Resembles de- Atractus cf. crassicaudatus DumeÂril and Bibron: pressiocellus in having anterior black ventral Dunn and Bailey, 1939: 8 (comments on a head pigmentation but differs in much smaller 26 AMERICAN MUSEUM NOVITATES NO. 3391 size, larger eye, low (vs. tall) supralabials be- length of prefrontal suture; prefrontals large, low eye, and narrow pale bars on a black longer than wide (greatest prefrontal width neck. Resembles darienensis in white su- 82% of greatest length); prefrontal suture 70% pralabials and a similarly positioned vague length of frontal plate; supraoculars large, lon- nuchal collar, but differs in having pale bars ger than wide; frontal slightly wider than on a black neck, a relatively larger eye, and long, roughly triangular in shape; interparietal a shorter but broader snout, with the rostral suture longer than prefrontal suture, 83% of plate more extensive dorsally. Resembles frontal length. hostilitractus in having pale bars or broken Eye small, contained 1.77 times in loreal ``rings'' on a black neck, but differs in hav- length, 2.8 times in snout length; eye length ing a longer, less rounded snout, loreal not about equal or marginally greater than dis- approaching internasal, white labials, and tance to lip; eyes set close to head, not pro- less extensive black pigmentation on under- truding out to level of edge of lips, hence not side of head and throat. visible in ventral view. Nasal divided above See Remarks below for further discussion and below just anterior to rear edge of naris, of differences from Atractus darienensis, its greatest length 70% of loreal length; lo- which seems to be the species most similar real long, 2.6 times longer than greatest to A. imperfectus, at least based on accessible height, not approaching internasal, entering characters. eye; no preoculars; supralabials 7, third and Dunn and Bailey (1939) likened this spec- fourth in eye, fourth wider than high; two imen to the Andean A. crassicaudatus, but it normal postoculars on left side, upper one is easily differentiated from that species and slightly larger, but three postoculars on right, I see no reason to suspect a close relationship lowermost a tiny scale; temporals 1 ϩ 2, the (see Remarks). upper one in second row elongated, extending past ends of parietals. PARTIAL DESCRIPTION Infralabials 7, ®rst pair in contact behind mental, ®rst four on each side in contact with Dunn and Bailey (1939) gave some scale a genial; a single pair of large genials, 2.5 counts for this specimen but curiously pro- times longer than wide; four large median vided nothing about the color pattern save gulars and preventrals between genials and mention of a ``black postocular streak across ®rst ventral. Head plate tubercles tiny, the last upper labial'' (but see below and ®gs. sparsely distributed, and inconspicuous. 9C, 10). Although the posterior color pattern, COLOR PATTERN OF HEAD AND NECK: The number of ventrals etc. are unknown, the fol- short section of neck is black, with two lowing description should allow it to be brownish white ringlike markings (1±2 scales matched whenever a second Panamanian wide) that are broken at the vertebral scale specimen is obtained. row (®g. 10). The ®rst pale ring (damaged by PROPORTIONS AND SCUTELLATION: The spec- a machete cut) is about 9 scales behind the imen fragment comprises 38 mm of head and parietals and about 7 scales anterior to the neck (®g. 10). An original total length of second; the two halves of the second ``ring'' about 350±400 mm seems likely, based on are offset by a distance of two scales in the comparisons with A. darienensis and A. hos- vertebral row. The top and upper sides of the tilitractus, whose heads are roughly similar in head are dark but, in bright light, are seen to size to that of A. imperfectus (table 1). Neck be of a slightly paler hue11 than the dark neck; scales smooth, without apical pits, in 17 rows. the slightly lighter aspect of the head is em- Head slightly wider than neck, which is a little phasized by a still paler, albeit vague, orang- wider than high, with rounded ventrolateral ish brown nuchal collar, which extends nar- edges; greatest head width 73.7% of head rowly and indistinctly across the ®rst nuchal length from tip of snout to end of parietals. Snout bluntly rounded in dorsal view, bluntly 11 It is not certain that the head would be paler than pointed in pro®le; rostral wider than high, the neck in life, since I have noticed differential fading well visible in dorsal view; internasals small, of black heads and bodies in some preserved snakes as wide as long, slightly more than half (52%) (e.g., Geophis). 2003 MYERS: SNAKES FROM PANAMA 27 scales and tips of the parietals, ending abrupt- tion of Colombian crassicaudatus.'' Com- ly at a lateral white area behind the mouth parison of a large series of BogotaÂ crassi- (®g. 9C). The supralabials are mostly white, caudatus in the American Museum fails to with excursions of dark head color forming a substantiate a close resemblance: For exam- serrated pattern above. An isolated blackish ple, the Colombian species differs in having mark occupies much of the last supralabial a larger eye relative to a shorter loreal and (®g. 9C), but does not convey the impression in having noticeably shorter genials (the of a ``black postocular streak'' as described main reason why the Panamanian specimen by Dunn and Bailey (1939). has 4 infralabials in contact with the genials, The underside of the head is mostly white, whereas Dunn and Bailey (loc. cit.) were with a noticeable suffusion of black on the `ùnable to ®nd a Colombian specimen with genials, mental, and adjacent edges of the in- more than three in contact''). fralabials (®g. 10). A median line of black I had originally hoped that this specimen markings commences on the preventrals, could be assigned to one of two other new posteriorly widening under the neck to cover Atractus described herein, namely darienen- most of each ventral. An irregular ventrolat- is, which has a similarly shaped head, vague eral white line, on the ends of the ventrals nuchal collar, and white labials, or hostili- and lower part of the ®rst scale row, extends tractus, which is super®cially similar in hav- posteriorly from behind the head and con- ing a ringed neck and extensive black pig- nects the pale dorsal rings. mentation under the throat. But I decided MAXILLO-PALATO-PTERYGOID ARCH: Ex- otherwise after direct comparison of snout amined in situ on right side. Maxilla arched, shapes (see ®g. 9), details of anterior color extending anteriorly past suture between ®rst patterns, eye size relative to loreal and dis- two supralabials, with 7 well-spaced re- tance to lip, and consideration of details of curved teeth, decreasing in size posteriorly. head plate sizes and proportions. I concluded First tooth springs from anterior tip of max- that the species represented by only a head illa; ®rst ®ve teeth large, with noticeable gaps and neck is distinctÐas also suspected by (not sockets) between 2 and 5, followed by Savage (1960: 32), who mentioned `àn un- a larger gap and two very small teeth; teeth described population [in Panama] allied to distally angular in cross section, with an an- Atractus crassicaudatus''. terolateral edge. Maxilla extending posteri- The closest resemblance seems to be with orly past small teeth as an elongate toothless Atractus darienensis, which differs from A. process. An expanded ¯ange on maxilla ex- imperfectus in a few details of coloration tending mediad and ventrad adjacent to the and, more convincingly, in proportions. The two small posterior teeth. Ectopterygoid ®rst two light brown interspaces (between forked, with one branch being an expanded black blotches) on the neck of darienensis ¯ange tightly bound against the expanded (®g. 1B) seem somewhat comparable to the maxillary ¯ange. Maxillary process of pala- pale broken ringlike markings (on black) in tine absent or weak. imperfectus (®g. 10). The anterior interspaces in darienensis are brown to the top of the REMARKS ®rst scale row, which is white like the anterior ventrals; the paler broken ``rings'' in im- Obtained in 1936, this partial specimen perfectus have some slight suffusions of pale was the ®rst Atractus collected in Panama. brown, but are mainly yellowish white like Dunn and Bailey (1939) compared it with the adjacent edges of the ventrals. The su- topotypes of Atractus crassicaudatus from pralabials in darienensis have little encroach- BogotaÂ, Colombia. They (1939: 8) stated ment of dark pigment from above, whereas that, except for a few minor differences (i.e., those of imperfectus are boldly serrated number of labials in contact with genials, rar- (compare ®g. 9A and 9C). ity in crassicaudatus of an elongated upper The holotypes of A. imperfectus and A. secondary temporal, and position of a post- darienensis seem roughly comparable in size, ocular marking), the ``characters of scalation with the latter having a slightly shorter head and markings are within the range of varia- that is, however, slightly wider across the tem- 28 AMERICAN MUSEUM NOVITATES NO. 3391 poral region than the head of imperfectus (ta- incomplete specimens gathered by Clark's ble 1). But direct comparison of the speci- Panama snake census: Sibynomorphus ni- mens shows imperfectus to have a relatively cholsi Dunn, 1933 (ϭ Dipsas nicholsi); Hy- broader snout in dorsal view, with this being dromorphus clarki Dunn, 1942; and Micru- re¯ected in proportions of the paired prefron- rus nigrocinctus yatesi Dunn, 1942 (ϭ M. tal plates, which are anteriorly tapered to a alleni yatesi). greater extent in darienensis (®g. 3B) than in imperfectus (®g. 3D). The width across the GENUS GEOPHIS posterior margin of the paired prefrontals is This genus was admirably revised by the same (4.3 mm) in both specimens, but the Downs (1967), who recognized four species anterior margin narrows to 2.0 mm in dari- in PanamaÐbrachycephalus, championi, enensis vs. 2.9 mm in imperfectus. Closer to godmani, and hoffmanni. Among subse- the snout tip, the width across the internasals quently described species were two from is 1.9 mm in darienensis vs. 2.1 mm in im- neighboring Costa Rica (G. downsi Savage, perfectus. The rostral plate extends more dor- 198l; G. talamancae Lips and Savage, 1994) sad in imperfectus than in darienensis. The and one from Colombia (G. betaniensis Res- somewhat smaller head of darienensis is trepo and Wright, 1987).12 slightly longer in snout length (table 1), which There has been confusion concerning the is re¯ected in length of the loreal plate (2.6 distribution of this genus in Panama, where mm in darienensis, 2.3 mm in imperfectus) there have been no good published records and also in relative eye size. The eye is rela- east of El Valle de AntoÂn in CocleÂ Province tively larger in imperfectus (36% of snout (G. hoffmanni). Reports of G. godmani and length, 56.5% of loreal length) than in dari- G. brachycephalus in central or even eastern enensis (30% of snout, 46% of loreal). The Panama are based on apparently mislabeled eye of imperfectus also is relatively larger specimens from the GML snake census and compared with its distance to the edge of the extrapolations therefrom. Nonetheless, as ev- lip (eye length/eye-to-lip distance ϭ 1.08 in idenced by a specimen reported herein, a imperfectus, 0.857 in darienensis). widely disjunct population of G. brachyce- All the differences elaborated above con- phalus does occur in the uplands east of the vince me that Atractus imperfectus is a spe- Panama Canal, and G. hoffmanni also occurs cies different from A. darienensis. Nonethe- there, as well in the central Panamanian low- less, one can scarcely predict the extent of lands and probably in northern Colombia. As variation in diagnostic characters derived also to be discussed, Geophis brachycephal- from two specimens from widely separated us s.l. seems to include an unrecognized sib- populations, particularly when one specimen ling species in western Panama. is only a fragment! If a whole specimen of One new species seemingly endemic to ''imperfectus'' ever turns up and disproves central Panama is described herein. The fol- this hypothesis, I would urge that the species lowing key includes the species now known name darienensis be given priority simply to occur in Panama. because that name is based on a complete holotype from a more precise type locality. KEY TO PANAMANIAN GEOPHIS In a ®rst draft of this description written a decade ago, I treated Atractus imperfectus as 1. Supraocular present, parietal plate not in contact with eye; tip of snout not noticeably a ``species inquirenda'', even stating in man- paler than rest of head ...... 2 uscript that it ``would seem poor judgement ± Supraocular absent, parietal bordering poster- to add a name to the literature without being odorsal part of eye; tip of snout whitish, able to describe the posterior color pattern contrasting with adjacent head scales .... and other missing details.'' But, considering ...... G. godmani the real rarity of this apparently diagnosable species, I now join my paleontological breth- 12 Another South American species (G. alasukai) was described by Gasc and Rodrigues (``1979'' [1980]) from ren in recognizing that specimens ofttimes French Guiana, but according to Hoogmoed (``1982'' are not perfect. Atractus imperfectus there- [1983]: 230) the name is a junior synonym of Atractus fore joins several other snakes named from ¯ammigerus. 2003 MYERS: SNAKES FROM PANAMA 29

2. Six supralabials, posterior temporal plate pre- The only Panamanian specimens of G. sent; body either uniformly dark or con- brachycephalus with de®nite locality seen by spicuously patterned above ...... 3 Downs (1967: 153) are from far western ± Five supralabials (a large postlabial does not border free edge of lip), temporals absent Panama. Central Panama was included on the (i.e., no scales positioned between ultimate basis of a series of specimens (ANSP 24723± supralabial and parietal); uniformly dark 24734) from ``Panama Sabanas''. These above except for light collar in juveniles specimens, consecutively numbered with a ...... G. hoffmanni partial specimen (ANSP 24723) of G. god- 3. At least posterior dorsal scales keeled; rostral mani (see below) having the same locality plate not posteriorly extended between in- data, are from the GML snake census (see ternasals; color variable ...... 4 Introduction). ± Dorsal scales smooth throughout; rostral posteriorly extended between internasals; dor- As summarized by Dunn (1949a: 47), the sum uniformly dark, venter weakly to GML ``sabanas collection'' was accumulated strongly banded ...... G. championi from various localities in the central Paci®c 4. Venter uniformly dark (glossy black in life) lowlands, ``from Capira to the west to Canita like dorsum; conspicuous white nape col- [CanÄita] in the mid-basin of the Bayano to lar; size small and slender (holotype, an the east''Ða straight-line distance of 120 adult male, 201 mm total length) ..... kmÐalthough ``the majority of the speci- ...... G. bellus mens come from the true `Panama sabanas' ± Venter pale, never uniformly dark; dorsum uniformly dark or with light blotches or lateral which lie between Panama City and Chepo.'' stripe; juveniles with or without a light collar; Dunn (1949a: 47, table 7) reported on over size larger (adults attaining maximum total 3900 snakes from this collection, with no lengths greater than 400 mm) ...... mention of Geophis in this or in any other ..... G. brachycephalus and G., species part of the GML lowland collection, even inquirenda though Dunn presented the ``sabanas'' Geo- phis to the Academy of Natural Sciences in RECORDS EXCLUDED FROM CENTRAL PANAMA 1942, seven years before publication of his Geophis brachycephalus (Cope): This is summary analysis. the most common Panamanian Geophis in The ``sabanas'' locality seems wrong on collections, but Downs' (1967: 146) state- the face of it. Neither G. brachycephalus nor ment that it ranges from ``Costa Rica south- the rare G. godmani are lowland savanna ward through Panama to Colombia'' needs snakes, and the thought that both might have correcting. South America was included in been found in sympatry in such an unlikely the range because of Downs' synonymizing habitat is hard to swallow and harder to di- of the Colombian Geophis nigroalbus, which gest. Furthermore, several specimens of the is shown herein to be distinct based on pro- little montane snake Trimetopon slevini portional and hemipenial characters (see (ANSP 24717±24719) also were presented Comparisons under Geophis bellus, new spe- by Dunn in 1942 and were cataloged as cies).13 ``Panama Sabanas''; neither were these mentioned in Dunn's 1949a paper. I surmise that 13 Some confusion concerning the range of Geophis Dunn either knew that the locality had to be brachycephalus may stem from my long delay in pub- lishing the present paper. I showed Downs the holotype wrong for these specimens and ignored them, of Geophis bellus sometime in the mid-1960s, which led or there was a cataloging error that he either him (Downs, 1967: 146, ftnote) to publish a statement was unaware of or never got around to cor- doubting that his inclusion of G. nigroalbus in the syn- recting. onymy of G. brachycephalus was justi®ed. Possibly en- Although the ``sabanas'' record for Geo- couraged by this, Restrepo and Wright (1987) resurrect- ed the Colombian G. nigroalbus without comment. In phis brachycephalus must be discarded as er- turn, Wilson et al. (1988: 416), followed by Lips and roneous, this species does occur in upland Savage (1994: 413±414), retained G. brachycephalus for wet forest in east-central Panama, as dem- Colombia and added G. nigroalbus for ``eastern Panama onstrated later in this paper. and Colombia''. The few Colombian specimens assigned by Downs (1967: 153) to G. brachycephalus are tenta- Geophis godmani Boulenger: This is a rare tively reassigned to G. nigroalbus (see footnote 16), a snake in Panama, and it evidently is a mon- species not known from Panama. tane species throughout its range. Downs 30 AMERICAN MUSEUM NOVITATES NO. 3391

level near community of Altos de Pacora (east of Cerro Jefe), Province of PanamaÂ, central Panama. The type locality is a few km northeastward of the summit of Cerro Jefe, upper RõÂo Pacora drainage, at roughly 9Њ15ЈN, 79Њ22ЈW. ETYMOLOGY: The speci®c name bellus is a Latin adjective meaning ``pretty'' and ``charming'', referring to the appearance of this elegant little snake in life. DIAGNOSIS: A small, slender member of the sieboldi group of Geophis distinguished from other members of the genus by the combination of (1) dorsal scales in 15 rows, Fig. 11. Geophis bellus, new species. The ho- smooth anteriorly but moderately keeled and lotype (KU 110703) shown approximately two striated on posterior part of body; (2) anterior times life size. temporal present or absent; (3) six supralabials; (4) eye small, going about 3.5±4 times (1967: 72), followed by Peters and Orejas- into snout length; (5) dorsal and ventral sur- Miranda (1970: 119), gave its distribution as faces uniformly dark (black in life) except ``Caribbean and Paci®c slopes of central for vivid white band across rear of head; and Costa Rica southward to the Canal Zone of (6) hemipenis unicapitate, about one-fourth Panama [at known localities] between 1300± bilobed, nearly acalyculate, with calyces con- 2100 meters above sea level.'' ®ned distally to lobes of the large capitulum. The three Panamanian specimens of G. The overall color pattern (®g. 11) alone godmani seen by Downs (1967: 75) are all distinguishes Geophis bellus from any other from the GML snake censusÐtwo from Fin- small colubrid in Panama. Other small Pan- ca LeÂrida in extreme western Panama, the amanian snakes having blackish or dark third (ANSP 24722) purportedly from the brown bodies, and heads partially or mostly ``Panama Sabanas'' in central Panama. This white, are Enuliophis sclateri, Ninia atrata, last specimen is cataloged consecutively with and Tantilla albicepsÐall of which have the series of G. brachycephalus and in prox- pale venters. See Comparisons for discussion imity with the series of Trimetopon slevini and contrast with relevant congeners (includ- discussed above, and the godmani record ing G. betaniensis, G. brachycephalus, G. ni- must be discarded for the same reasons. The groalbus, and G. talamancae). most likely correct locality for the ``sabanas'' Geophis brachycephalus, G. godmani, and DESCRIPTION OF HOLOTYPE Trimetopon slevini is Finca LeÂrida in Chiri- quõÂ Province, which provided some census Despite its small size and slender habitus, material and where all three species occur.14 the male holotype is unambiguously sexually mature because (1) the hemipenial spines had hardened, (2) the relatively large testes Geophis bellus, new species (about as long as the snake's head) are Figures 11±14, 15C; map 2 strongly tubular, (3) the slender vasa defer- HOLOTYPE: KU 110703 (®eld no. CWM entia are convoluted, and (4) sperm were pre- 3223), an adult male caught by C.W. Myers sent in a small fragment of testis examined on December 13, 1964, at 700 m above sea under a compound microscope. The anterior (right) testis is 6.8 mm long and the posterior 14 After reading the above discussion in manuscript, J. is 6.5 mm long, both being about 2.1 mm M. Savage kindly consulted Dunn's notes in his posses- wide as measured in situ. sion and concurs that the ``sabanas'' records of Geophis ROPORTIONS AND CUTELLATION and Trimetopon must be a cataloging error. Dunn's notes P S :Itisa contain no reference to specimens of either genus as very small, short-tailed snake, 201 mm total having been part of the composite ``sabanas'' collection. length, 32 mm tail length (tail 15.9% of to- 2003 MYERS: SNAKES FROM PANAMA 31

bluntly pointed, projecting well beyond lower jaw; rostral well visible from above, where its length is about 23% of its distance to frontal, not projecting between internasals; internasals small, slightly wider than long, less than half as long as prefrontal suture; prefrontal suture 64% length of frontal plate; frontal broader than long, roughly rhomboidal in shape (strongly angular anteriorly); parietals the largest head plates, their median suture shorter than frontal plate; supraocular forming posterior two-thirds of upper eye margin. Nasal deeply grooved above naris, divided below, anterior and posterior parts subequal, their combined lengths four-®fths that of loreal; long loreal entering eye; no preocular; eye very small, contained about 4 times in snout (left eye 3.6ϫ, right eye 4.1ϫ), its length (and height) smaller than its distance from lip; supralabials 6, third and fourth in eye on left side, fourth only on right side; ®fth supralabial largest and in contact with parietal on left side only; one postocular, higher than long, subequal in size to supraocular; temporals 0 ϩ 1 on left side, 1 ϩ 1 on right (®rst temporal the apparent result of a division of supralabial 5). Chin tapered, anteriorly rounded; mental rounded, wider than long, separated from genials by ®rst pair of infralabials, which are in medial contact; infralabials 7, pairs 1±4 in Fig. 12. Head scutellation of Geophis bellus, contact with anterior genials, 4±5 with pos- new species (holotype), ϫ6. Fifth supralabial and terior genials; posterior genials small and posterior temporal shown in gray in lateral views. posteriorly rounded, in contact with each other; short posterior intergenial suture 33% the length of anterior intergenial suture; two me- tal). Body quite slender, barely wider than dian gulars and one preventral (gular wider high (at maximum of about 4.8 ϫ 4.6 mm), than long) between posterior genials and ®rst rounded ventrolaterally. Dorsal scales in 15 ventral. Tubercles sparsely present on some rows throughout, smooth on anterior half of dorsal and ventral head plates, but these pre- body, with rows 4±12 becoming moderately sumed sensory organs are minute and incon- keeled and striated on posterior half; inconspicuous. spicuous anal knobs present on lower lateral COLOR PATTERN: Color in life uniform scales above cloaca; apical pits not detected glossy black above and below, with a con- at 40ϫ magni®cation under re¯ected light. spicuous white band across rear of head (®g. One preventral (undivided gular), 131 ven- 11). The white band anteriorly includes the trals, undivided anal plate, 33 pairs of sub- posterior halves of supralabials 4 and the rear caudals. edges of the frontal plate, encroaching slight- Following description of head (®g. 12) af- ly on supraocular and postocular scales; the ter Downs' format (1967, for G. brachyce- band extends posteriorly onto the nape for a phalus); see also table 2 for measurements: distance of two scales behind the parietal Head barely distinct from neck; snout long, plates; the rear lateral edges of the band an- 32 AMERICAN MUSEUM NOVITATES NO. 3391

TABLE 2 Measurements (in mm) of an Adult Male Geophis bellus, New Species, and Juvenile Males of Geophis brachycephalus and Geophis negroalbus

gle anteroventrally under the head and ex- Following account based mainly on left tend to the posterior genials as a pair of whit- everted organ, except for description of the ish streaks. After some 20 years in alcohol lobes which is based on right organ15: Mod- the black parts of the dorsum and venter have erately bifurcate (®g. 14), with lobes com- faded to dark brown, and the front part of the prising about one-fourth of total length. Dis- head has differentially faded to a lighter tinctly capitate, but on asulcate side the nude brown than the body. collar is almost interrupted medially by MAXILLO-PALATO-PTERYGOID ARCH: Right spines; capitulum comprising half the length maxilla (®g. 13) extending slightly anterior of organ on asulcate side but nearly 70% on to the suture between second and third supralabials (equal to anterior extension of pal- 15 The hemipenes were only half everted at time of atine), bearing 10 rather stout, subequal preservation. Both were subsequently dissected out, teeth; anterior tip of maxilla pointed, tooth- soaked in glycerin, then in a saturated solution of tri- less; posterior end of maxilla with slight ven- sodium phosphate, and in¯ated with carmine-dyed petroleum jelly. The right organ everted fully but is some- tral curve, tapering to blunt point; maxilla what twisted, and many of the spines are evidently bent somewhat dorsoventrally compressed, bear- out of normal alignment. The left organ is straight and ing large palatine process. No maxillary pro- the spines are aligned more symmetrically, but owing to cess on palatine. Anterior end of ectoptery- small tears the lobes could only be partially everted. The above description therefore is based mainly on goid single, not expanded. the left hemipenis, with recourse to the right organ for HEMIPENIS: Fully everted right hemipenis the distal parts of the lobes. The ®gure likewise is a about 6.5 mm long, reaching to subcaudal 6. composite. 2003 MYERS: SNAKES FROM PANAMA 33

Fig. 13. Right maxilla of Geophis bellus, new species (holotype), shown in lateral and ventral views. sulcate side. Sulcus spermaticus divides below midpoint of hemipenis but well onto capitulum, where the branches diverge and distally curve slightly around to other side (a modi®ed centrifugal con®guration if the Fig. 14. Hemipenis of Geophis bellus, new lobes are not abnormally twisted), terminat- species. Everted organ of holotype shown in sul- ing close to the apices of lobes. cate and asulcate views, ϫ10. The drawing is a Organ entirely spinose (but spines distad composite, based on left hemipenis except that the from base of capitulum rather ¯exible, seem- lobes are added from right organ (see footnote ing less or not calci®ed): Proximal part of 15). The relatively large capitulum is nearly aca- hemipenis with spinules and small spines, lyculate, with weak calyces being con®ned to the then two fairly large spines that are widely small lobes. separated, one on each side of sulcus spermaticus. Midsection with a few dozen mod- unexpanded) distal end of the ectopterygoid erate-sized hooked spines that are most con- process and in maxillary features. Pertinent centrated on asulcate side. Capitulum rather maxillary characters include a small number sparsely covered with small spines (or spine- of subequal teeth, dorsal and ventral ¯atten- like papillae, see above); in lateral view, the ing of the maxilla, a toothless anterior pro- spines proximal and lateral to the sulcus jection, and a ventrally curved, bluntly point- branches are seen to be aligned in oblique ed posterior end (compare ®g. 13 with that rows, but between the branches of the sulcus of G. brachycephalus [Downs, 1967: 26]). spermaticus and on the lobes the arrange- The presence of an anterior temporal plateÐ ment is one of vague horizontal rows; tips of on the right side of the holotype of G. bel- lobes spinose. Capitulum acalyculate, except lusÐis a rare condition in the sieboldi group. for indication of weak calyces on lobes, these It seems quite clearly to have arisen by di- calyces being formed by elongated low tissue vision of the ®fth supralabial (compare lat- ridges connecting bases of some spines. An eral views of head in ®g. 12), thus providing elongated lateral naked pocket on basal a clue to the method of original loss of the fourth of organ; from the perspective of the primary temporal in the sieboldi group (see asulcate side, the nude pocket is on the left- Downs, 1967: 14). hand side of the left organ and the right-hand The single specimen of Geophis bellus side of the right organ. was shown to F. L. Downs shortly after he had completed his monograph of the genus. COMPARISONS It is the specimen mentioned in a footnote Geophis bellus is a member of Downs' (Downs, 1967: 146) as leading him ``to (1967: 137) sieboldi species group, with doubt that my inclusion of G. nigroalbus which it generally agrees in features of scu- Boulenger in the synonymy of G. brachy- tellation, small eye, long projecting snout, cephalus is justi®ed.'' Downs (1967: 152) and, especially, in the simple (unforked, had not seen the Colombian type of Boulen- 34 AMERICAN MUSEUM NOVITATES NO. 3391 ger's nigroalbus, but direct comparison of the holotypes of bellus and nigroalbus con- vinces me that they are different species (see below). I also closely compared the type of bellus with the only central Panamanian specimen of brachycephalus, which probably was collected within a few kilometers of the former. Measurements of these specimens are given in table 2, with further comparisons following. Geophis bellus is a much smaller snake than the Colombian G. nigroalbus, in which males attain a total length of at least 370 mm, and the slender bellus is relatively less robust. The juvenile holotype of G. nigroalbus, although 67 mm shorter than the adult holotype of G. bellus, nonetheless has a slightly larger head (®gs. 15A, 16, and table 2). The holotype of nigroalbus resembles bellus in having a dark body and a similarly positioned and vivid pale band across the head and nape (absent or vestigial in other specimens of nigroalbus seen), but differs in the uniformly pale venter (adult nigroalbus may acquire dark transverse banding ventrally).16 Fig. 15. Heads of Geophis, all ϫ4.5. A. Col- Although the adult male holotype of G. ombian holotype of G. nigroalbus, juvenile male bellus is essentially identical in total length 134 mm total length (see ®g. 16 for whole body). to the nearly sympatric juvenile male of G. B. Central Panamanian specimen of G. brachy- brachycephalus, the latter specimen (®gs. cephalus (GML), juvenile male 202 mm total length (see ®g. 18 for whole body). C. Central 15B, 18) is a more robust snake with a longer Panamanian holotype of G. bellus, new species, and wider head and noticeably larger eye; an adult male 201 mm total length (pale lateral streak behind head band is re¯ected light; body 16 Unfortunately, I have seen too few Colombian spec- uniform black above and below, as seen in ®g. imens to be able to fully characterize Geophis nigroal- 11). bus, or even to conclude that but a single species is represented. The juvenile holotype (134 mm total length) from Pavas in Valle de Cauca is the most dis- most of the various head plate and suture tinctive of ®ve specimens because of its broad, well- demarcated nape band. A smaller juvenile (116 mm to- measurements are smaller in bellus, except tal), from Santa Rita in Antioquia, has a less extensive, for slightly longer internasals and parietals. poorly de®ned nape band that is dorsally incomplete, The central Panamanian specimen of brachy- and this specimen has a relatively larger eye (but onto- cephalus also differs from the probably sym- genetic change in relative eye size is not uncommon). Three adults, one with a faint trace of a nape band, differ patric bellus in having a pale-spotted dorsum from the two juveniles in having dark bands across the and a pale venter (®g. 18). ventrals. Comparisons of the holotype of G. bellus The holotype of G. nigroalbus and three other Col- with specimens of G. brachycephalus from ombian specimens (BMNH 98.10.27.3, FMNH 43727, western Panama lead to similar conclusions: 54882) examined by me are listed in Downs (1967: 153) under the name brachycephalus. The ®fth specimen ex- Geophis bellus is a much smaller, more slen- amined is LACM 136675, one of a series reported by der species. The striking nape band of bellus Restrepo and Wright (1987) from Betania in Valle de seems to be a retention of a variably present- Cauca. The specimens with data are from roughly 1500± or-absent juvenile feature in brachycephalus 1700 m in the Cordillera Occidental, but Downs also listed one (not seen by me) from the Cordillera Oriental and nigroalbus. Downs (1967: 150) noted (LandaÂzuri [900 m ®de Medem, 1965: 342], Depto. San- that a pale collar, if present, is usually lost in tander). brachycephalus by 150 mm SVL (adult ho- 2003 MYERS: SNAKES FROM PANAMA 35

Fig. 16. Colombian holotype of Geophis nigroalbus Boulenger. BMNH 1946.1.6.50 (formerly 1906.4.30.71), a juvenile male 134 mm total length, shown ϫ1.7. lotype of bellus ϭ 169 mm SVL), with traces ported specimen) of G. talamancae,is11 of the collar only occasionally persisting to mm longer than the adult male specimen of a larger size. According to Downs, brachy- bellus, but it obviously is a more robust cephalus attains a maximum total length of snake, with a much stockier body and a rel- 418 mm in males and 460 mm in females. atively larger head (greatest body width in I also compared the holotype of Geophis talamancae ϭ 7.2 mm vs. 4.8 mm in bellus; bellus directly with the Costa Rican holotype greatest head width ϫ head length to end of of G. talamancae Lips and Savage (1994), parietals in talamancae ϭ 6.0 ϫ 8.5 mm vs. which is similar in size, relative tail length, 4.5 ϫ 7.0 mm in bellus). Geophis talaman- and in numbers of ventrals and subcaudals.17 cae appears to be close to G. brachycephalus The juvenile female holotype (the only re- and perhaps G. nigroalbus. Extent of keeling probably does not distinguish it from bra- 17 The holotype of Geophis talamancae is now cataloged as LACM 147196. My own scale counts and mea- chycephalus as thought by Lips and Savage surements for this specimen include 133½ ventrals, 32 (1994: 411), but the ventral color pattern pairs of subcaudals, 212 mm total length, 33 mm tail (dark pigmentation across bases of ventrals length (15.6% of total). It is a juvenile female (said to causing a fuzzy pattern of alternating dark be an adult in the original description), with inactive ovaries containing a few tiny ova and with ¯at, noncon- and white cross-bands) distinguishes tala- voluted oviducts. mancae at least from Panamanian G. brachy- 36 AMERICAN MUSEUM NOVITATES NO. 3391

lected in sympatry with the similar-sized G. nigroalbus. It is a distinctive species, separated from nigroalbus at a glance by dark ventrolateral stripes that sharply con®ne and set off the pale midventral area. Geophis betaniensis may be more closely related to G. nigroalbus than initially thought. Restrepo and Wright (1987: 191) provisionally assigned G. betaniensis to the championi species group, but Lips and Savage (1994: 414) later moved it to the sieboldi group based on ``a complete concordance'' in head shape and scutellation. Cursory examination of the paratype of G. betaniensis shows that it lacks certain derived maxillary characteristics shared by most members of the sieboldi group; the maxilla is not ventrolaterally compressed and lacks a toothless anterior projection (the ®rst tooth springs from the anterior tip of the bone). Nonetheless, a Costa Rican member of the sieboldi groupÐGeophis ze- ledoniÐalso lacks a ``distinct toothless tip'' on the maxilla, which is dorsoventrally ¯at- tened only along its posterior half (®de Downs, 1967: 145, 175). But the maxilla of Geophis betaniensis is relatively high and I see no sign of dorsoventral compression even Fig. 17. Hemipenis of Geophis nigroalbus posteriorly (right maxilla examined in situ). Boulenger. Left organ of FMNH 43727 from Co- Tentatively accepting Lips and Savage's lombia (no other data) shown left to right in sul- (1994: 414) reassignment of betaniensis (as cate and asulcate view, ϫ5.9. a species with presumably plesiomorphic maxillary features) makes the sieboldi group cephalus and Geophis, species inquirenda. of Geophis the only one known from South This type of ventral pattern (suggested in one America. case to mimic noxious millipedes [Leonard Geophis betaniensis was described as hav- and Stebbins, 1999]) is approached in some ing ``a small tentacle-like projection from the specimens of the Colombian G. nigroalbus posterior margin of the nostrils'' (Restrepo (FMNH 54882, LACM 136675). Although I and Wright, 1987: 191±192, ®g. 3). A pho- do not suggest conspeci®city, it needs noting tograph shows the ``hair-like `tentacle''' on that G. nigroalbus is not distinguished from the right side of the holotype. A tentacle was talamancae by separation of postocular and said to be present only in the left nostril of supraocular by an anterior projection of the the paratype (LACM 136189), but I could parietal (Lips and Savage, 1994: 411).18 see no sign of it when I examined the spec- Finally, a word on the Colombian Geophis imen in early 2002. My own guess is that betaniensis Restrepo and Wright (1987), these structures, initially present in three of which is known from two specimens col- four nostrils, were foreign to the two snakesÐperhaps defensive arthropod urticat- 18 This character comes from Restrepo and Wright ing hairs that might glance off the snakes' (1987: 195), who evidently observed the condition in head scales but temporarily lodge in their one population of Geophis nigroalbus (I have seen nostrils. LACM 136675 from their sample; the parietal extends to the eye above exceedingly small postoculars), but the It is of course impossible to determine at condition does not pertain to the holotype nor to the few this time whether there is signi®cant color other nigroalbus that I have seen (footnote 16). and size variation in Geophis bellus, but its 2003 MYERS: SNAKES FROM PANAMA 37 speci®c status, as ®rst suggested by the striking coloration and small adult size, is strongly corroborated by hemipenial differences: In G. bellus the hemipenis (®g. 14) is bilobed for about a fourth of its length, with weak calyces only on the distal lobes; the large capitulum otherwise is sparsely covered with small spines (or large papillae). In G. nigroalbus (®g. 17) and G. brachycephalus (®g. 20) the organ is barely or not bilobate and the entire capitulum is covered with well-developed calyces.

REMARKS I found the unique specimen of Geophis bellus at night, while walking on a muddy road through cut-over evergreen seasonal forest, at 700 m above sea level. The type locality is on the ``Piedras-Pacora Ridge'', the low continental divide that separates the watershed of the RõÂo Pacora (Paci®c versant) from that of the upper RõÂo Chagres (Atlan- tic). Geophis bellus shares the Piedras-Pacora Ridge with G. brachycephalus and G. hoffmanni (see below), and also with the new species Atractus depressiocellus and A. imperfectus of this paper. The one new species of Geophis and two new Atractus are known Fig. 18. Geophis brachycephalus (Cope), a only from single specimens collected over a juvenile male from the Piedras-Pacora Ridge in central Panama (GML specimen, 202 mm total period of 29 years. The two previously length; see ®g. 15B for lateral view of head). named species of Geophis seem to be just as rare on the Piedras-Pacora Ridge (although common in other parts of their ranges), since weak and disappearing toward the end of the each is known there from a single specimen, tail; however, the ®ne striations of the dorsal as follows. caudal scales continue up to the terminal spine. Supraocular present; no preocular; Geophis brachycephalus (Cope) one postocular; 0 ϩ 1 temporals; six su- Figures 15B, 18, 20A; map 2 pralabials (3±4 in eye); seven infralabials ARANGE EXTENSION: I assign to Geophis (1±4 touching anterior genials, 4±5 touch- brachycephalus a GML specimen from ing posterior genials); 138 ventrals, undivid- ``Cerro Azul'' on the Piedras-Pacora Ridge ed anal, 50 pairs of subcaudals; see table 2 in central Panama; it was collected for the for measurements. The dorsal color is blue Gorgas Memorial laboratory by A. Herrera in alcohol, with 11 pale dorsal rings or in October 1965. It is a juvenile male with paired spots on the posterior body and sev- uncalci®ed hemipenial spines, 202 mm in eral on the tail (®g. 18). total length. It has 15±15±15 rows of dorsal As remarked under Atractus depressiocel- scales that are smooth on the neck, becom- lus, the ``Cerro Azul'' of GML ®eld parties ing striated and weakly keeled by ventral 13 is equivalent to Cerro Jefe, the 980 m high and then progressively more strongly keeled point on the Piedras-Pacora Ridge about 37 posteriorly; the posterior half of the body km NE of downtown Panama City. Speci- and proximal half of the tail are quite mens so labeled might have been taken any- strongly keeled, with the keels becoming where from about 5 km southwestward of 38 AMERICAN MUSEUM NOVITATES NO. 3391

Cerro Jefe to about 5 km to the north, most Geophis, species inquirenda likely in an elevational range of about 200± [G. brachycephalus, auctorum] 800 m. Figures 19, 20B The distribution of Geophis brachyce- MATERIAL: Bocas del Toro: About 3 km phalus is partly clari®ed earlier in this paper W Almirante, 40 m (KU 110701); La Loma (under Records Excluded from Central Pan- (MCZ 19325±19326); RõÂo Changena, 830 m ama). The present specimen is the only one (KU 110702); RõÂo Changena (FMNH known to me from central Panama. It ap- 130969). ChiriquõÂ: Upper RõÂo ChiriquõÂ, For- pears to mark the eastern and southern limit tuna Dam Site, 1000 m (AMNH 114317± of the range, where brachycephalus seems 114319); south slope Quebrada de Arena, very rare compared with its abundance in 1120 m (AMNH 124015). the highlands of Costa Rica and western Panama. As already mentioned, South COMMENTARY American records of this species are tentatively assigned to the Andean Geophis ni- The earliest specimens, from La Loma in groalbus. Bocas del Toro, were collected and identi®ed as Geophis brachycephalus by Dunn (1942: 4). Downs (1967: 154) had no additional TAXONOMIC COMMENT specimens from Bocas del Toro, but he did mention a few small Costa Rican samples Downs (1967: 149) commented that ``The lacking blotches. I have not seen these, but I color pattern is the most perplexing variable have collected and examined additional Pan- in G. brachycephalus, and is at least partly amanian specimens that may be conspeci®c responsible for the lengthy synonymy''. with the La Loma specimens and distinct Some populations appear to be highly poly- from brachycephalus. morphic in color pattern, as exempli®ed in These specimens are uniformly black or Panama by Slevin's (1942: 474±476) de- brown in preservative (not bluish as in some scription of a large population sample from albeit not all preserved G. brachycephalus) the vicinity of Boquete in the ChiriquõÂ high- and all lack pale spots or partial ringlike lands: markings (a vague pale nape band present or absent). The hemipenis of an Atlantic-ver- 26 specimens show reddish spots, blotches or short sant specimen differs from that of a highland stripes, specimen of G. brachycephalus in having a 23 specimens are uniform in color, relatively long, slender capitulum with no 2 specimens show a white collar and no lateral spots, trace of bilobation (®g. 20). However, part of 2 specimens show a white collar and red lateral spots. the difference (especially relative slender- Such populations, as well as the single spec- nessÐbut not relative lengthÐof the capitulum) conceivably might be an artifact of imen (®g. 18) from central Panama, seem to eversion technique (see comment under ®t in with the original descriptions of Co- Methods of Study). lobognathus [ϭ Geophis] brachycephalus Several of the newer specimens are from and its simultaneously described synonym low and intermediate elevations (40±830 m) C. dolichocephalus (Cope, 1871: 211±212). in Bocas del Toro, but four are from a col- However, I strongly suspect that some lection that I made in the highland valley of snakes with uniformly dark dorsa from the RõÂo ChiriquõÂ in ChiriquõÂ Province (1000± along the Atlantic versant of western Pana- 1120 m). This valley may be considered as ma (and Costa Rica?) are currently mas- delimiting the eastern end of the Cordillera querading in collections under the name de Talamanca and the western end of the Ser- Geophis brachycephalus. Although I am not ranõÂa de TabasaraÂ. This high Paci®c-side val- prepared to give adequate attention to this ley is in¯uenced by moisture-laden air spill- problem, it may be helpful to call attention ing southward over the low continental di- to it by segregating the possible sibling spe- vide, providing access to animals and plants cies as follows. from the Caribbean versant (the herpetofauna 2003 MYERS: SNAKES FROM PANAMA 39

TABLE 3 Comparison of Segmental Counts between the Boquete Population of Geophis brachycephalus and Combined Samples of Geophis, species inquirenda in Western Panamaa

also has a strong Talamancan component, but are made for somewhat dissimilar methods only two of 50 species collected have pri- of counting ventrals.19 marily Paci®c-drainage distributions [Myers Downs (1967: 148) remarked that ventral and Duellman, 1982: 14]). Unfortunately, the counts are somewhat lower in Panamanian Geophis collected in the valley of the upper G. brachycephalus from Finca LeÂrida to the ChiriquõÂ are all females, and their assignment west of Boquete, and that both these Pana- to the Atlantic-versant population is based on manian populations have considerably lower segmental counts and absence of pale spots counts than in Costa Rican populations. The or partial rings in all four specimens. ranges and means for all his Costa Rican There is a pronounced difference in num- specimens combined were given as 131±148 bers of ventrals and subcaudals between Pan- (138.5) in 54 males and 135±145 (140.7) for amanian populations with color polymor- 70 females. These ranges encompass those phism (brachycephalus) and those apparently for species inquirenda in Panama, but little without (species inquirenda). For purposes more can be said without detailed examina- of comparison, I calculated population statis- tion of the situation in Costa Rica, from tics from the counts given by Slevin (1942: where considerably more material is now 474±476) for his large collection of Geophis available. See Remarks below. brachycephalus from the vicinity of Boquete, COLOR IN LIFE where it was the most abundant snake in the cafetals. As shown in table 3, Boquete spec- In life, the dorsa are uniformly black; preserved specimens (MCZ 19325±19326, KU imens of brachycephalus have signi®cantly higher numbers of ventrals and subcaudals in 19 There is, as noted under Methods of Study, a dif- both sexes. There is no overlap in female ference in methods of counting ventrals, with Slevin's ventral counts. Overlap in male ventrals and counts expected to average probably one or a few ven- subcaudals is caused by one brachycephalus trals more. Downs (1967: 148) recounted Slevin's material and determined sex of some specimens (e.g., ju- (CAS 78983) with exceptionally high counts veniles) left unsexed by Slevin. Downs, whose counting of 142 ϩ 46, without which the Boquete method was closer to the one I used, gave the following ranges would be 121±129 ventrals and 37± ranges and means for the Boquete sample of Geophis 41 subcaudals. The high ventral count, how- brachycephalus: 34 males, 119±140 (mean, 124.9), 28 females, 123±132 (mean, 127.1). Adding the ``preven- ever, was con®rmed by Downs (1967: 148), trals'' to my counts for Geophis, species inquirenda who gave 140 ventrals for this specimen. The yields 2 males, 134±140 (mean, 137.0 Ϯ 4.24), 7 fe- differences are maintained when adjustments males, 138±144 (mean, 140.4 Ϯ 2.07). 40 AMERICAN MUSEUM NOVITATES NO. 3391

somewhat bluish because it was about to shed.

NATURAL HISTORY The habitat is lowland and lower montane rain forest and cloud forest in an elevational range of 40±1120 m. One (KU 110702) was under a log by day. I found a lowland specimen (KU 110701) as it was crawling on a log across a rocky forest steam, at night in the rain. The highest elevation specimen Fig. 19. Geophis, species inquirenda? A spec- (AMNH 124015) was active at night, on the imen (AMNH 114317&) collected in 1976 near ground in cloud forest, at the edge of a small the Fortuna Dam Site (prior to construction), 1000 seepage pool in which Hyla graceae was m, upper valley of the RõÂo ChiriquõÂ (ChiriquõÂ calling. Three specimens (AMNH 114317± Province, western Panama). Although it is Paci®c 114319) from the area of the Fortuna Dam drainage, the highland valley has a climate com- site were found on a second-growth hillside; parable to that of the Atlantic versant. Pending a workman found one, I caught another at further investigation, four female Geophis from night on a trail near the river, and I found the area are tentatively associated with unicolored the third by dayÐon the ground near a specimens from the Atlantic versant of western rocky stream, at about 10 a.m. on a sunny Panama. morning.

110701 [see below]) may fade to brown with REMARKS the head becoming lighter than the body. A satisfactory resolution of the taxonomic Two specimens from the RõÂo Changena problem presented by this species inquirenda (FMNH 130609, KU 110702) in preservative is outside the scope of the present paper, have pale but vague nape bands, which may which I had intended to con®ne to problems have been more pronounced in life (grayish in eastern Panama. Costa Rican specimens white in KU 110702). My ®eld descriptions and names currently in the synonymy of for several specimens follow: Geophis brachycephalus will need to be con- KU 110701: Shining black above and on tip of chin sidered. There especially needs to be exam- and infralabials. Venter immaculate white except sub- ination of more hemipenes to determine if caudals heavily marked with black. [Specimen has faded to brown in preservative, with the head becom- the differences seen in ®gure 20 are corre- ing paler than the body.] lated with presence or absence of color-pat- AMNH 114317±114319 [see ®g. 19]: Uniform shin- tern polymorphism. ing black above [one bluish, see below]; lower, posterior edge each scale white in scale row 1. Venter greenish white, turning black on chin and under tail. Geophis hoffmanni (W. Peters) Iris dark, appears black. Tongue pale gray to base of Map 2 fork, tips of fork unpigmented. AMNH 124015: Black above (with slight grayish Downs (1967: 155) gave the range of this white tinge on lower side of neck behind jaw), white species as ``low and moderate elevations in below, turning grayish black under tail. Iris dark. Honduras . . . and Nicaragua, southward Tongue pale gray becoming white (unpigmented) on along both Caribbean and Paci®c versants of tips of fork. Costa Rica into Panama''. This little snake W.E. Duellman noted one to be gray dorsally evidently is common in parts of Costa Rica and subcaudallyÐbut this specimen (KU and has long been known to occur in extreme 110702) is black in preservative and proba- western Panama; Downs' easternmost station bly was preparing to shedÐwith a creamy in Panama was El Valle de AntoÂn in CocleÂ white belly and a grayish white nuchal collar. Province, based on AMNH 76016; a second Similarly, I noted that a specimen in the se- specimen (KU 116902) has subsequently ries AMNH 114317±114319 above was been collected there. El Valle de AntoÂn, an 2003 MYERS: SNAKES FROM PANAMA 41

Fig. 20. Hemipenes of Geophis. Right-side organs shown at same scale (ϫ6.7) in sulcate and asulcate view. A. Geophis brachycephalus, from a highland population with pattern polymorphism (UMMZ 57958; Boquete, western Panama). B. Geophis, species inquirenda, from Atlantic-versant population apparently lacking pattern polymorphism (La Loma, western Panama; MCZ 19326).

old volcanic crater, is shown as the western- As discussed elsewhere (Cadle and Myers, most station in map 2. 2003), As listed below, two additional specimens The Madden Forest Preserve (formerly in the Canal extend the range eastward in Panama from Zone, now in Prov. PanamaÂ) straddles Madden Road, El Valle to the central lowlands and thence a paved highway between Summit and Madden Dam. to the Piedras-Pacora Ridge (map 2), and a The roughly rectangular Preserve, about 3 ϫ 5 km, third specimen indicates the occurrence of G. lies across the very low continental divide, which hoffmanni in Colombia. All three specimens swings close to the Paci®c coast and seldom exceeds 200 m above sea level on this part of the isthmus. are dark brown above, whitish below, with The northerly (Atlantic) drainage via the upper RõÂo dorsal scales smooth or mostly smooth, in Chilibre system ¯ows into the Chagres below Mad- 15±15±15 rows, supralabials 5 (3±4 in eye); den Dam; the southerly (Paci®c) drainage adds to the temporals absent (see Remarks); infralabials Canal waters via the RõÂo Pedro Miguel. The vegeta- tion is lowland monsoon rain forest. 6 (1±3 in contact with ®rst genials). MADDEN FOREST: The late Howard W. The surprising thing about this specimen is Campbell collected a specimen (AMNH that it is the only veri®able specimen of Geo- 113561) of Geophis hoffmanni in the former phis from the central lowlands, which, how- Canal Zone in July 1968. It was found on ever, are not as well known as might be ex- ``Madden Forest Rd. 20 yds. N George pected, nor is the lowland fauna as uniformly Green Pk.'' It is a male 200 mm in total distributed as might be expected (for a recent length, including 29 mm (14.5%) tail length, contribution, see IbaÂnÄez et al., ``1995'' with 125 ventrals and 29 pairs of subcaudals. [1997]). The poorly collected Madden Forest In addition to the presence of anal ridges, the has several snakes that are not found on the dorsal scales are weakly keeled at the end of well-known Barro Colorado Island only the body and are striated on the rear of the about 24 km to the northwestward, including body and tail. Dipsas nicholsi and Dipsas viguieri (Cadle 42 AMERICAN MUSEUM NOVITATES NO. 3391 and Myers, 2003), as well as Geophis hoff- The specimen is a female 189 mm in total manni. length, including 23 mm (12.2%) tail length; PIEDRAS-PACORA RIDGE: A specimen the dorsal scales are weakly keeled above the (FMNH 152049) of Geophis hoffmanni was tail and the dorsal caudal scales are striated; collected by Harold Trapido and Emmett ventrals 132; subcaudals in 28 pairs. The Reid Dunn on May 3, 1952, at 2200 ft [670 snout seems unusually acuminate compared m] on ``Cerro La Victoria''. Trapido20 was at with eight Panamanian specimens, and the that time a staff scientist at the Gorgas Me- parietal plate is in contact with the eye be- morial Laboratory and Dunn was a visiting tween a reduced supraocular and postocular. scientist. As expounded below under Re- But the supraocular and postocular also are marks, Cerro La Victoria was an important separated by the parietal in occasional Cen- GML ®eld station that I place in the Piedras- tral American specimens (Downs, 1967: Pacora ridge system a few km south of Cerro 157). Therefore, the Colombian snake is ten- Jefe (map 2). tatively assigned to Geophis hoffmanni, with The specimen is a female 125 mm in total which it agrees in most particulars. length, including 15 mm (12%) tail length; the smooth dorsal scales become striated (but REMARKS not keeled) at the posterior end of the body; ventrals 125; subcaudals in 23 pairs. As discussed by Downs (1967: 159), the COLOMBIA: A single specimen (AMNH reduction to ®ve supralabials in Geophis 108373) is available from Colombia. It was hoffmanni appears to be the result not of one of a several specimens with disassociated scale fusion but of a shortening of the jaw. data sent as a gift by J.R. Tamsitt from the A postlabial (i.e., the ®rst scale not bordering Universidad de Los Andes in BogotaÂ, being the free edge of the lip) appears to be a ho- the ``smaller, pointed-head snake'' that has mologue of the sixth supralabial in related only ``Colombia'' for locality data (Tamsitt, species. And a scale dorsal to the postlabial in letter to R.G. Zweifel, June 8, 1965). It is and ventral to the posterior part of the pari- cataloged as having been collected by James etal appears to represent the original poste- R. Tamsitt and Dario Valdivieso in 1960± rior temporalÐtemporals being pragmatical- 1963. These workers collected too widely in ly de®ned simply as scales positioned mainly Colombia for an itinerary to be informative or completely between the parietals and pos- (e.g., see locality list in Valdivieso and Tam- terior supralabial(s). sitt, 1963), although the snake presumably The easternmost record of Geophis hoff- came from northern Colombia. manni in Panama is in the Piedras-Pacora ridge system (map 2) and is of particular in- 20 Harold Trapido (1916±1991) started his professional terest in the context of the present paper. Of life as a herpetologist but ended it as an Emeritus Pro- the snakes from the eastern half of Panama fessor of Medical Entomology. His name appeared on about 20 herpetological publications, but army service that are treated herein, ®ve of the eight spe- in the Sanitary Corps starting in 1942 led to his assign- cies occur in that relatively small area. The ment in 1944 to the Gorgas Memorial Laboratory, where following remarks document the locality for he was involved in the early testing of DDT for malarial Geophis hoffmanni, the most widely occur- control. About 360 herpetological specimens collected ring of the ®ve species. during his military service in Panama (1944±1945) were given to the American Museum. After the event of several fatal human cas- ``Trap'' afterwards served as Biologist on the GML es of sylvan yellow fever in 1948±1949, the staff from 1946 to 1956, and made signi®cant contri- Gorgas Memorial Laboratory established butions to the epidemiology of yellow fever in Central temporary ®eld stations in the forest at Cerro America. But he continued to collect amphibians and reptiles on the side, ostensibly ``to get good pictures in La Victoria in PanamaÂ Province east of Pan- color of important species'' (Clark, 1952: 20), and he ama City. Primary objectives were to collect sent valuable Central American material to the Field mosquitoes simultaneously from the forest Museum (Resetar and Voris, 1997: 503), including about ¯oor and from platforms up into the upper 1200 specimens from Panama. He also helped arrange for the Field Museum to receive selected material from canopy, and to survey arboreal forest mam- the Panamanian snake census conducted by Clark at mals in order to discover natural reservoirs GML (footnote 4). of the yellow fever virus. 2003 MYERS: SNAKES FROM PANAMA 43

Cerro La Victoria is not shown on stan- in the general vicinity of Cerro Tobova dard maps nor is it listed in Fairchild and (9Њ12ЈN, 79Њ23ЈW).21 Handley's (1966) gazetteer of GML collec- Therefore, the northern end of Cerro La tion sites, but it was an important GML col- Victoria on the Paci®c versant is only several lecting area during the early 1950s. I know km south of the better known Cerro Jefe in Cerro La Victoria only through the publica- the Atlantic versant (shown as `Èl Jefe'' in tions of Galindo et al. (1950, 1951, 1956), the Madden Lake Watershed inset in the old the ®rst two of which contain outline maps 1:100,000 map Canal Zone and Vicinity pub- showing a series of collecting stations lished by the U.S. Army Map Service). Cerro aligned northwest of the Paci®c lowland Jefe lies just to the north and higher than the town of Pacora, about midway between Pa- continental divide, at 9Њ14ЈN, 79Њ23ЈW, with cora and the mouth of the upper RõÂo Chagres an elevation of 3254 ft (982 m) according to at Madden Lake. the aforesaid map and as con®rmed by my altimeter reading of 980 m (Myers, 1969: The series of stations lettered from ``B'' to ``D'' was located along a woods road, recently built, which 27). From the summit of Cerro Jefe, during gives access to a ridge now called Cerro La Victoria, the occasional clear period, the Cerro La Vic- between the Rio Juan Diaz to the west and the Rio toria region (including Cerro Tobova) and Cabra to the east, while to the north it is limited by the Paci®c coastal plain are visible to the the drainage of the Rio Cascades, a tributary of the upper Rio Chagres. These three stations are all on the south. side of the ridge draining to the Rio Cabra... Before Trapido and Dunn's specimen of Geophis the advent of the road . . . this region was loosely hoffmanni, from 2200 ft. (670 m), apparently spoken of as ``Cerro Azul'' and appears with that des- was collected above station D near the crest ignation in various botanical and zoological publica- of Cerro La Victoria (Galindo et al., 1950: tions. The natives of the region now commonly re- serve the term ``Cerro Azul'' for the ridges to the east ®g. 7 [aerial photo showing location of sta- of the Rio Cabra, while Cerro La Victoria rises to the tion D]). The specimen was said to have been west of this river . . . found `ìn bromeliads'' (presumably ground The summit of the ridge at Cerro La Victoria is bromeliads) with a curious assemblage of 2300 feet [700 m] above sea-level and is exposed to 22 far more wind than the slopes of the mountain. The other small reptiles. Half a century ago, the more exposed places along the summit have a cover Paci®c-versant forest in the immediate region of low trees not exceeding 30 feet in height. Station was already being encroached by road build- D was located at 2100 feet [640 m] and about 200 ing and slash-and-burn agriculture (Galindo feet below the summit, in a ravine with taller trees. et al, 1950: ®gs. 2±3, 7±8), and the forest by (Galindo et al., 1950: 536±537) now has either disappeared or reverted to Photographs included an aerial view ``from second growth. above the forested slopes of Cerro La Vic- toria [looking] down the valley of the Rio MUSEUM ABBREVIATIONS AND Cabra'', and magni®cently high forest with ACKNOWLEDGMENTS `à covering canopy 80 to over 100 feet ABBREVIATIONS: The following collection high'' (Galindo et al., 1950: ®gs. 2, 5). The abbreviations are used in this paper: lower elevation forest at elevations of 400± 1200 ft (120±365 m) was `ìntermediate in AMNH American Museum of Natural History, appearance between deciduous forest and reptile collection, New York rain-forest'', with a marked dry season (Gal- ANSP Academy of Natural Sciences of Phil- indo et al., 1956: 543). adelphia, Philadelphia From the above information, it is evident that Cerro La Victoria is a high, forested 21 The map elevation for the neighboring summit of Cerro Tobova is 2430 ft (740 m), only 40 m higher than ridge above the west side of the south-¯ow- given for Galindo et al. (1950) for Cerro La Victoria. ing RõÂo Cabra, which drains the southern 22 Alan Resetar (personal commun.) could ®nd no ex- slopes of the Piedras-Pacora Ridge (i.e., the plicit information in Trapido's notes, but observed that continental divide) to the west of the head- the FMNH catalogue has ``several other specimens with the same collection event information (same locality, waters of the RõÂo Pacora. It is an area of same date, `ìn bromeliads'') . . . Lepidophyma ¯avi- premontane rain forest. The La Victoria maculatum, Gymnophthalmus speciosus, and Ninia ma- Ridge connects to the main divide probably culata''. 44 AMERICAN MUSEUM NOVITATES NO. 3391

BMNH The Natural History Museum, London searched FMNH ®les for needed data. Igna- CAS California Academy of Sciences, San cio De la Riva provided photographs that al- Francisco lowed correction of the troublesome identi- FMNH Field Museum of Natural History, Chi- ®cation in footnote 2. JuliaÂn Faivovich cor- cago rected the Resumen, and John E. Cadle and GML Gorgas Memorial Laboratory, Panama KU Museum of Natural History, University Jay M. Savage kindly made time to read and of Kansas, Lawrence comment on the complete text. I am indebted LACM Natural History Museum of Los An- to all. geles County, Los Angeles MCZ Museum of Comparative Zoology, rep- REFERENCES tile collection, Harvard University, Cambridge Adler, Kraig. 1989. Herpetologists of the past. In K. Adler (editor), Contributions to the history ACKNOWLEDGMENTS: My early work in of herpetology. Society for the Study of Am- Panama was supported by National Institutes phibians and Reptiles, Contributions to Herpe- of Health Grant No. GM-12020 to the Uni- tology 5: 5±141. Barbour, Thomas. 1925. A new frog and a new versity of Kansas (W.E. Duellman and C.W. snake from Panama. Occasional Papers of the Myers, co-investigators), with logistics and Boston Society of Natural History 5: 155±156. of®ce space provided by the Gorgas Memo- Cadle, John E. 1984. Molecular systematics of rial Laboratory in Panama. Neotropical xenodontine snakes: II. Central My wife Joan aided this work in many American xenodontines. Herpetologica 40(1): ways, and she and our children Charles and 21±30. Tracy suffered my long absences from Pan- Cadle, John E., and Charles W. Myers. 2003. Sys- ama City with good-natured resourcefulness. tematics of snakes referred to Dipsas variegata My 3-year term (1964±1967) as Visiting Sci- in Panama and Western South America, with entist at the Gorgas Memorial Laboratory revalidation of two species and notes on defensive behavior in the Dipsadini (Colubridae). was facilitated by my colleagues there, in- American Museum Novitates, in press. cluding especially Abdiel J. Ademas, Gra- [Clark, Herbert C.] 1937. Progress report on snake ham B. Fairchild, Pedro Galindo, Carl M. census (1929±1936). In Annual Report of the Johnson, Eustorgio MeÂndez, and Martin D. Gorgas Memorial Laboratory, 1936 (75th Con- Young. Subsequent work in Panama was also gress, 1st session, House Document 46): 15± expedited by the Gorgas Memorial Labora- 18. tory, and supported by the American Muse- Clark, Herbert C. 1942. Venomous snakes. Some um of Natural History and a variety of other Central American records. Incidence of snake sources. A. Stanley Rand, of the Smithsonian bite accidents. The American Journal of Trop- Tropical Research Institute, was repeatedly ical Medicine 22(1): 37±49. [Clark, Herbert C.] 1952. Snake census of Pana- more helpful than he remembers. ma. In Annual Report of the Gorgas Memorial For the past loan of comparative material Laboratory, 1951 (82nd Congress, 2nd session, from their respective collectionsÐand their House Document 278): 20±21. patienceÐI remain grateful to the following Commission, Atlantic-Paci®c Interoceanic Study. colleagues: Edwin N. Arnold, Colin J. Mc- 1971. Interoceanic Canal Studies 1970. Wash- Carthy, and Andrew F. Stimson (BMNH); the ington, DC: U.S. Government Printing Of®ce, late James E. BoÈhlke and Edmond V. Mal- 129 pp. ϩ annexes. nate (ANSP); William E. Duellman (KU); Cope, Edward Drinker. 1871. Ninth contribution Robert F. Inger, Hymen Marx, and Alan Re- to the herpetology of tropical America. Pro- setar (FMNH); David A. Kizirian and Kent ceedings of the Academy of Natural Sciences Beaman (LACM); Arnold G. Kluge and of Philadelphia 1871: 200±224. Cunha, Osvaldo Rodrigues da, and Francisco Pai- Gregory E. Schneider (UMMZ); Eustorgio va do Nascimento. 1983. OfõÂdios da AmazoÃnia MeÂndez (GML); JoseÂ P. Rosado and the late XXÐas espeÂcies de Atractus Wagler, 1828, na Ernest E. Williams (MCZ). AmazoÃnia oriental e MaranhaÄo. (Ophidia, Co- I bene®tted from conversations about lubridae). Boletim do Museu Paraense EmõÂlio Atractus and other snakes with Frances J. Goeldi, Nova SeÂrie, Zoologia 123: 1±38 ϩ 2 Irish and Samuel B. McDowell. Alan Resetar pls. 2003 MYERS: SNAKES FROM PANAMA 45

Daniel, H[ermano]. 1949. Las serpientes en Co- Trapido. 1951. Ecological observations on for- lombia. Revista Facultad Nacional de Agron- est mosquitoes of an endemic yellow fever area omia de MedellõÂn 9(36): 301±333. in Panama. The American Journal of Tropical Dowling, Herndon G. 1951. A proposed standard Medicine 31(1): 98±137. system of counting ventrals in snakes. British Galindo, Pedro, Harold Trapido, Stanley J. Car- Journal of Herpetology 1(5): 97±99. penter, and Franklin S. Blanton. 1956. The Downs, Floyd Leslie. 1967. Intrageneric relation- abundance cycles of arboreal mosquitoes dur- ships among colubrid snakes of the genus Geo- ing six years at a sylvan yellow fever locality phis Wagler. Miscellaneous Publications, Mu- in Panama. Annals of the Entomological Soci- seum of Zoology University of Michigan 131: ety of America 49(6): 543±547. iv, 1±193. Gasc, Jean-Pierre, and Miguel T. Rodrigues. Duke, James A., and Duncan M. Porter. 1970. ``1979'' [1980]. Sur la preÂsence du genre Geo- Bioenvironmental and radiological-safety fea- phis (Colubridae, Serpentes) dans la reÂgion sibility studies, Atlantic-Paci®c interoceanic ca- guyanaise. Bulletin du MuseÂum National nal. DarieÂn phytosociological dictionary. Co- d'Histoire Naturelle, 4e seÂrie, vol. 1, section A, lumbus, OH: Battelle Memorial Institute, [iv] no. 4: 1121±1130. ϩ 70 pp. Greene, Harry W., and Roy W. McDiarmid. 1981. Dunn, Emmett Reid. 1933. A new snake from Coral snake mimicry: does it occur? Science Panama. Copeia 1933(4): 193±194. 213: 1207±1212. Dunn, Emmett Reid. 1942. New or noteworthy Hoogmoed, Marinus S. 1980. Revision of the ge- snakes from Panama. Notulae Naturae of the nus Atractus in Surinam, with the resurrection Academy of Natural Sciences of Philadelphia of two species (Colubridae, Reptilia). Notes on 108: 1±8. the herpetofauna of Surinam VII. Zoologische Dunn, Emmett Reid. 1944. Notes on Colombian Verhandelingen 175: 1±47 ϩ 6 pls. herpetology, II. Caldasia 2(9): 397±405. Hoogmoed, Marinus S. ``1982'' [1983]. Snakes of Dunn, Emmett Reid. 1949a. Relative abundance the Guianan region. Memorias do Instituto Bu- of some Panamanian snakes. Ecology 30(1): tantan 46: 219±254. 39±57. IbaÂnÄez D., Roberto, Fernando A. Arosemena, Dunn, Emmett Reid. 1949b. The application of Frank A. SolõÂs, and CeÂsar A. Jaramillo. ``1994'' Fisher's formula to collections of Panamanian [1995]. An®bios y reptiles de la SerranõÂa Pie- snakes. Ecology 30(4): 533±536. dras-Pacora, Parque Nacional Chagres. Scientia Dunn, Emmett Reid, and Joseph R. Bailey. 1939. (PanamaÂ) 9(1): 17±31. Snakes from the uplands of the Canal Zone and IbaÂnÄez D., Roberto, CeÂsar A. Jaramillo, Marianela of Darien. Bulletin of the Museum of Compar- ArrunaÂtegui, Querube Fuenmayor, and Frank ative Zoology 86(1): 1±22. A. SolõÂs. ``1995'' [1997]. Inventario bioloÂgico Fairchild, Graham B., and Charles O. Handley, Jr. del Canal de PanamaÂ. Estudio herpetoloÂgico. In 1966. Gazetteer of collecting localities in Pan- V.H. Tejera, R. IbaÂnÄez D., and G. Arosemena ama. In R.L. Wenzel and V.J. Tipton (editors), G. (editors), El inventario bioloÂgico del Canal Ectoparasites of Panama: 9±20 ϩ foldout map. de PanamaÂ. II. Estudio ornitoloÂgico, herpeto- Chicago: Field Museum of Natural History. loÂgico y mastozooloÂgico: 107±110 (executive Fernandes, Ronaldo. ``1995'' [1996]. Variation summaries) ϩ 111±159. Scientia (PanamaÂ), and taxonomy of the Atractus reticulatus com- 1995, nuÂmero especial 2 [published in March plex (Serpentes: Colubridae). ComunicacËoÄes do 1997] Museu de CieÃncias e Tecnologia da PUCRS. King, Wayne. 1959. Vertebra duplication, an os- SeÂrie Zoologia 8: 37±53. teological anomaly widespread in snakes. Her- Fernandes, Ronaldo, Eliza Maria Xavier Freire, petologica 15(2): 87±88. and Giuseppe Puorto. 2000. Geographic varia- Leonard, William P., and Robert C. Stebbins. tion of the Brazilian Atlantic rain forest snake 1999. Observations of antipredator tactics of Atractus maculatus (GuÈnther, 1858) with the re- the sharp-tailed snake (Contia tenuis). North- validation of Rhabdosoma zebrinum Jan, 1862 western Naturalist 80: 74±77. (Serpentes: Colubridae). Boletim do Museu Na- Lips, Karen R., and Jay M. Savage. 1994. A new cional. Nova SeÂrie, ZoologõÂa 419: 1±8. fossorial snake of the genus Geophis (Reptilia: Galindo, Pedro, Harold Trapido, and Stanley J. Serpentes: Colubridae) from the Cordillera de Carpenter. 1950. Observations on diurnal forest Talamanca of Costa Rica. Proceedings of the mosquitoes in relation to sylvan yellow fever Biological Society of Washington 107(2): 410± in Panama. The American Journal of Tropical 416. Medicine 30(4): 533±574. Martins, M., and M.E. Oliveira. 1993. The snakes Galindo, Pedro, Stanley J. Carpenter, and Harold of the genus Atractus Wagler (Reptilia: Squa- 46 AMERICAN MUSEUM NOVITATES NO. 3391

mata: Colubridae) from the Manaus region, Museum of Zoology University of Michigan central Amazonia, Brazil. Zoologische Mede- 114: 1±224. delingen 67: 21±40. Peters, James A., and Braulio Orejas-Miranda. Marx, Hymen. 1960. A new colubrid snake of the 1970. Catalogue of the Neotropical Squamata genus Atractus. Fieldiana, Zoology 39(38): Part 1. Snakes. U.S. National Museum Bulletin 411±413. 297: viii, 1±347. Medem, Federico. 1965. BibliografõÂa comentada Prado, Alcides. 1944. Serpentes da Colombia, de reptiles colombianos. Revista de la Acade- com a descricËaÄo de duas novas espeÂcies de mia Colombiana de Ciencias Exactas, Fisicas y Atractus. Ciencia (Mexico City) 5(4±5): 111. Naturales 12(47): 299±346. Prado, Alcides. 1945. Um novo Atractus da Co- Myers, Charles W. 1969. The ecological geogra- lombia. Ciencia (Mexico City) 6(2): 61. phy of cloud forest in Panama. American Mu- Prado, Alcides. 1946. Notas o®oloÂgicas 19. Atrac- seum Novitates 2396: 1±52. tus da ColoÃmbia, com a redescricËaÄo de treÃs no- Myers, Charles W. 1974. The systematics of vas espeÂcies. Memorias do Instituto Butantan Rhadinaea (Colubridae), a genus of New World 18(1944±1945): 109±111 ϩ pl. snakes. Bulletin of the American Museum of Rand, A. Stanley, and Charles W. Myers. 1990. Natural History 153(1): 1±262. The herpetofauna of Barro Colorado Island, Myers, Charles W. 1982. Blunt-headed vine Panama: an ecological summary. In A.H. Gen- snakes (Imantodes) in Panama, including a new try (editor), Four Neotropical rainforests: 386± species and other revisionary notes. American 409. New Haven, CT: Yale University Press. Museum Novitates 2738: 1±50. Resetar, Alan, and Harold K. Voris. 1997. Her- Myers, Charles W. 1986. An enigmatic new snake petology at the Field Museum of Natural His- from the Peruvian Andes, with notes on the Xe- tory, Chicago: the ®rst one hundred years. In nodontini (Colubridae: Xenodontinae). Ameri- T.W. Pietsch and W.D. Anderson, Jr. (editors), can Museum Novitates 2853: 1±12. Collection building in ichthyology and herpe- Myers, Charles W., and John E. Cadle. 1994. A tology. American Society of Ichthyologists and new genus for South American snakes related Herpetologists, Special Publication 3: 495±516. to Rhadinaea obtusa Cope (Colubridae) and Restrepo T., Jorge H., and John W. Wright. 1987. resurrection of Taeniophallus Cope for the A new species of the colubrid snake genus ``Rhadinaea'' brevirostris group. American Geophis from Colombia. Journal of Herpetol- Museum Novitates 3102: 1±33. ogy 21(3): 191±196. Myers, Charles W., and John E. Cadle. [MS]. On Roze, Janis A. 1961. El geÂnero Atractus (Serpen- the snake hemipenis, with notes on Psomophis tes: Colubridae) en Venezuela. Acta Biologica and techniques of eversion: a response to Venezuelica 3(7): 103±119. Dowling. Herpetological Review, in press. Roze, Janis A. 1966. La taxonomõÂa y zoogeogra- Myers, Charles W., and William E. Duellman. fõÂa de los o®dios en Venezuela. Caracas: Univ- 1982. A new species of Hyla from Cerro Col- ersidad Central de Venezuela, 362 pp. orado, and other tree frog records and geo- Sanchez-C., Hernan, Olga CastanÄo-M., and Glad- graphical notes from western Panama. Ameri- ys Cardenas-A. 1987. Diversidad de los reptiles can Museum Novitates 2752: 1±32. in Colombia. In J. Orlando Rangel Ch. (editor), Myers, Charles W., and A. Stanley Rand. 1969. Colombia diversidad biotica I: 277±325. Bo- Checklist of amphibians and reptiles of Barro gota: Universidad Nacional de Colombia. Colorado Island, Panama, with comments on Savage, Jay M. 1960. A revision of the Ecuador- faunal change and sampling. Smithsonian Con- ian snakes of the colubrid genus Atractus. Mis- tributions to Zoology 10: 1±11. cellaneous Publications, Museum of Zoology PeÂrez-Santos, Carlos, and VõÂctor MartõÂnez. 1997. University of Michigan 112: 1±86. Serpientes del Parque Nacional de Coiba (Pan- Savage, Jay M. 1981. A new species of the se- amaÂ). In S. Castroviejo (editor), Flora y fauna cretive colubrid snake genus Geophis from del Parque Nacional de Coiba (PanamaÂ) Ma- Costa Rica. Copeia 1981(3): 549±553. drid: 445±456. Agencia EspanÄola de Coopera- Schmidt, Karl P., and D. Dwight Davis. 1941. cioÂn Internacional. Field book of snakes of the United States and PeÂrez-Santos, Carlos, and Ana G. Moreno. 1988. Canada. New York: G. P. Putnam's, frontisp. ϩ O®dios de Colombia. Museo Regionale di xiii, 365 pp. Scienze Naturali (Torino), monogra®e 6, 517 Slevin, Joseph R. 1942. Notes on a collection of pp. reptiles from Boquete, Panama, with the de- Peters, James A. 1960. The snakes of the subfam- scription of a new species of Hydromorphus. ily Dipsadinae. Miscellaneous Publications, Proceedings of the California Academy of Sci- 2003 MYERS: SNAKES FROM PANAMA 47

ences, 4th series, 23(32): 463±480 ϩ pls. 39± Wilson, Larry David, James R. McCranie, and 42. Kenneth L. Williams. 1998. A new species of Smith, Hobart M. 1958. Handlist of the snakes of Geophis of the sieboldi group (Reptilia: Squa- Panama. Herpetologica 14(4): 222±224. mata: Colubridae) from northern Honduras. Terry, Robert A. 1956. A geological reconnais- Proceedings of the Biological Society of Wash- sance of Panama. Occasional Papers of the Cal- ington 111(2): 410±417. ifornia Academy of Sciences 23: 1±91 ϩ 4 Wright, Willard H. 1970. 40 years of tropical foldout maps. medicine research. A history of the Gorgas Me- morial Institute of Tropical and Preventive Torres de ArauÂz, Reina. 1970. Bioenvironmental Medicine, Inc. and the Gorgas Memorial Lab- and radiological-safety feasibility studies, At- oratory. Washington, DC: Gorgas Memorial In- lantic-Paci®c interoceanic canal. Human ecol- stitute, xii ϩ 426 pp., frontisp., 18 pls. ogy of route 17 (SasardõÂ-MortõÂ) region, DarieÂn, Zaher, Hussam. 1999. Hemipenial morphology of Panama. Columbus, OH: Battelle Memorial In- the South American xenodontine snakes, with stitute, [ii] ϩ 172 ϩ [xviii] pp. a proposal for a monophyletic Xenodontinae Valdivieso, Dario, and James R. Tamsitt. 1963. and a reappraisal of colubroid hemipenes. Bul- Records and observations on Colombian rep- letin of the American Museum of Natural His- tiles. Herpetologica 19(1): 28±39. tory 240: 1±168. Recent issues of the Novitates may be purchased from the Museum. Lists of back issues of the Novitates and Bulletin published during the last ®ve years are available at World Wide Web site http://library.amnh.org. Or address mail orders to: American Museum of Natural History Library, Central Park West at 79th St., New York, NY 10024. TEL: (212) 769-5545. FAX: (212) 769- 5009. E-MAIL: [email protected]

a This paper meets the requirements of ANSI/NISO Z39.48-1992 (Permanence of Paper).