Plica pansticta (Myers and Donnelly) Myers & Donnelly, 2008

Myers, C. W. & Donnelly, M. A., 2008, The Summit Herpetofauna Of Auyantepui, Venezuela: Report From The Robert G. Goelet American Museum-Terramar Expedition, Bulletin of the American Museum of Natural History 2008 (308), pp. 1-147 : 107-121

Status of Anolis eewi Roze

This nominal taxon was described by Roze (1958b) on the basis of a specimen (fig. 61) from the Chimantá massif. The specific name eewi is a noun in the genitive case derived from the initial letters of Ernest E. Williams, late authority on anoline lizards. Roze compared the specimen only with Anolis fuscoauratus , but Vanzolini and Williams (1970: 85) examined the holotype of eewi and effectively synonymized the name by stating that the specimen ‘‘falls within the variation of [ Anolis chrysolepis ] planiceps ’’.

Gorzula (1992: 275) worked extensively on Chimanta and revived eewi as a subspecies of Anolis chrysolepis . More recently, Gorzula and Señaris (1999: 144–145) resurrected Anolis eewi as a valid species for upland and highland populations in the Venezuelan Guayana, restricting A. chrysolepis planiceps (as A. n. nitens ) to lowland localities. According to these authors, although Anolis eewi shares many general characteristics with A. c. planiceps ,

A. eewi can be readily distinguished by its shorter

legs. A regression was made for tibia length

against SVL … using data from seven specimens

…. The sample included males and females (SVL

from 33.6 to 69.4 mm). The value for a tibia

length of 16.1 mm at 65 mm body length is well

below the ranges given by Vanzolini and

Williams (1970) for male (17.8 to 23.3 mm)

and female (17.6 to 21.1 mm) A. nitens [chryso-

lepis]. (Gorzula and Señaris, 1999: 144–145) Their regression analysis combined seven males and females from several localities (regression statistics are not given); there was no mention of comparison with any nearby lowland population. Our own small samples support Gorzula and Señaris’ (1999) claim that tepui specimens have relatively short tibias. However, this is not inconsistent with a pattern of discordant geographic variation in tibia length as analyzed by Vanzolini and Williams (1970) for Anolis chrysolepis throughout its range.

Regressions of tibia length on SVL for four males and three females from the summit of Auyantepui are shown in figure 62A, B. The slopes of the regressions (0.27 8, 0.25 ♀) for the Auyantepui data are within the lower ends of the ranges given by Vanzolini and Williams (1979: tables 119, 121) for geographic samples of male and female Anolis chrysolepis ; estimated tibia lengths at 65 mm SVL (17.5 mm 8, 16.9 mm ♀) are only slightly lower (by 0.3 and 0.7 mm) than the ranges given by Vanzolini and Williams (1970: tables 120, 122) for these estimates. Gorzula and Señaris’ (1999: 145) estimation of 16.1 mm of tibia length, for a hypothetical 8/ ♀ lizard of 65 mm SVL, falls close to the regression lines for either males or females from Auyantepui, as does the female holotype of Anolis eewi (52 mm SVL, 14.5 mm tibia).

To enlarge sample size, the four males from Auyantepui were combined with four males from Cerro Guanay, a locality that also contributed specimens to the sample used by Gorzula and Señaris. Our Auyantepui + Cerro Guanay sample yielded a regression (fig. 62C) similar to that for Auyantepui alone. Gorzula and Señaris’ estimated 16.1 mm of tibia length for 65 mm SVL would fall slightly outside the 95 % confidence limits, but well within the 95 % prediction limits for new observations.

Unfortunately, no series was available from lowlands adjacent to these tepuis. A juvenile male (AMNH R-103759) collected by Myers at a lowland site about 60 km south of the Río Orinoco and 200 km north of Auyantepui fits the tepui data marginally better than it does the northern coastal data; the measurements are 40.6 mm SVL, 11.6 mm tibia length. The estimating equation for figure 62C predicts a tibia length of 11.0 mm, compared with 12.3–12.5 mm for a 40.6 mm male lizard from northern coastal Venezuela. A single specimen naturally can fit any number of slopes and provides little information in a statistical sense, but this one example suggests that lowland specimens from south of the Río Orinoco might not be unambiguously distinguished from tepui specimens on the basis of tibia length.

Figure 62D shows a regression for males from a lowland region far to the south of Auyantepui and Cerro Guanay. The sample (AMNH specimens) comes from collections made at and near the lowland base camps of the Tapirapecó (Río Mavaca) and Cerro Neblina expeditions. This is a steeper slope than obtained from the tepuis, but the reason may reflect geographic distance rather than elevation. A single specimen (MBUVC 7044 ♀) caught at 1270 m elevation on Cerro Tamacuari— 100 km SE of the Tapirapecó base camp (Myers and Donnelly, 1997: 4– 5)—has a tibial length of 20.8 mm at 74 mm SVL. This large female (Myers and Donnelly, 1997: fig. 45) would fit well if the regression for lowland males in figure 62D could be accurately extrapolated past the last data point, but its much larger size precludes proper comparison.

Vanzolini and Williams (1970: 22) had access only to about a dozen Venezuelan specimens from the region south of the Río Orinoco. Although more individual specimens now reside in collections, there remains a lack of good samples for regression analysis. Nonetheless, the tepui data do show lizards of shorter legs when compared with Vanzolini and Williams’ (1970: 16 [map B], 27) major samples designated as Falcón and Northeastern Venezuela —comprised of localities paralleling the Caribbean coast well north of the Río Orinoco. Excluding sexual dimorphism, Anolis chrysolepis from this northern region seems relatively homogeneous in tibia length (regression coefficients 5 0.33–0.34 8, 0.28– 29 ♀; estimated tibia lengths at 65 mm SVL 5 20.5–20.7 mm 8, 19.4–19.5 mm ♀).

Tibia lengths in our small tepui samples from Auyantepui and Cerro Guanay are most similar to those in Vanzolini and Williams’ (1970: tables 119–122) major geographic samples designated as ‘‘Tapirapés’’, some 2400 km southeast of Auyantepui in northeastern Mato Grosso, and ‘‘Villavicencio’’, at the eastern foot of the Andes roughly 1300 km southeastward from Auyantepui. These two widely separated major samples of chrysolepis have regression coefficients (0.23–0.27 8, 0.24–0.26 ♀) very similar to those shown in table 8 for the tepui samples; estimated tibia lengths at 65 mm SVL (17.8– 18.4 mm 8, 17.6–18.0 mm ♀) are not appreciably different from the tepui estimates in table 8. Vanzolini and Williams (1970: 31–33, 58, 237–238 [graphs 32, 34]) included four Venezuelan males from Cerro Duida (a tepui) and nearby Esmeralda in their subsamples designated as the Colombo-Guianan transect and the Venezuelo-Brasilian transect, and concluded that ‘‘it is clear that the low Duida values are due to the influence of Villavicencio … not of Tapirapés’’.

It appears from the analysis by Vanzolini and Williams, and from data herein, that populations of relatively short-legged Anolis chrysolepis occur from the Andes eastward well into eastern Bolívar State. It would be interesting to know whether there are elevational correlates in tibia length, but, in any case, variation in tibia length is discordant with dewlap coloration—the key character for the pragmatic subspecies arrangement of Vanzolini and Williams (1970) and Ávila- Pires (1995). Dewlaps of male and female lizards in Venezuela south of the Río Orinoco show some intrapopulational variation but are basically red to orange with some dark and light scales, thus agreeing with Anolis chrysolepis planiceps . Ávila-Pires (1995: 70) listed eight scale characteristics of this subspecies (under the name A. n. nitens ). Our tepui material generally agrees in these characters, with occasional exceptions of

TABLE 8

Regression Statistics for Tibia Length on Snout-Vent Length a

in Venezuelan Samples of Anolis chrysolepis

individual variation, most commonly in the ‘‘double row of slightly enlarged vertebrals distinct from nape to base of tail’’; a transition between this double row and adjacent dorsals often obscures the distinction at midbody, and several specimens lack a distinct double row even on the nape.

Anolis chrysolepis as recognized is a polytypic species of great geographic range and considerable variational complexity. It would be difficult to segregate the Venezuelan Guayana portion as a separate species without some such evidence as sympatry or convincing molecular data. Unless such evidence becomes available, the name Anolis eewi Roze should remain in the synonymy of Anolis chrysolepis planiceps . Although this widespread lizard tends to be most common in lowland forest, it has a wide elevational range and is one of the most frequent lowland invaders of tepuis, where populational densities generally appear to be relatively low.

FAMILY TROPIDURIDAE

Tropidurus bogerti Roze Figures 63–66

Tropidurus bogerti Roze, 1958a: 247–250 , figs. 1–4. Holotype AMNH R-61011 from Auyantepui, [1100 m?; see Remarks], Estado Bolívar, Venezuela, collected by W. H. Phelps in January 1938. Myers, 1997: 4, fig. 1. Gorzula and Señaris, 1999: 155, photos 103–104.

MATERIAL: Camp 1, 1700 m: AMNH R-140218. Camp 2, 1750 m: AMNH R- 140219, EBRG 2921–2924. Camp 3, 1850 m: AMNH R-140220–140227, 140228 (eggs), EBRG 2925–2928, 2929 (eggs). Camp 5, 2100 m: AMNH R-140229, EBRG 2930– 2932. All from the 1994 AMNH –TERRAMAR Expedition to Auyantepui. ADDITIONAL MA- TERIAL: Auyantepui, 1100 m (3609 ft.) [elevation may be in error, see Remarks], AMNH R-61011 (holotype). Auyantepui Plateau, 2200 m (7218 ft.): AMNH R-61012, 61014– 61015, 61017 (paratypes). Type specimens from the 1937–1938 AMNH –Phelps Venezuelan Expedition. (Another paratype, AMNH R-61016, was exchanged to the MCZ in 1958; four additional paratypes 5 MBUCV 3053–3056 fide Roze, 1958a).

Based on dissection only of the AMNH share of specimens (N 5 11) from the 1994 expedition, females are appreciably smaller than the males. Five females, all reproductively active (see Natural History below), average 70.2 mm (67–75 mm) SVL, whereas four adult males average 88.3 mm (80–92 mm) SVL. Two additional specimens are immature males of 56 mm and 74 SVL; the last individual ( AMNH R-140218) has faint indications of black patches starting to appear under the thighs and in the cloacal area.

COLORATION: In life (fig. 63), overall black above, with light orange spots that are formed from small clusters of orange granular

scales; white or tan scales are interspersed among the orange spots in smaller clusters of scales or as individual scales. Ventral surfaces spotted or mottled bluish white or pale blue on black (ventral pattern in fig. 64). Venter in juveniles bronzy white with vague black reticulum. Adult males have a large black patch under each thigh and smaller pre- and postcloacal patches. Tongue pinkish gray or whitish, with gray tip. Iris turquoise blue, except noted as green in one adult male.

HEMIPENIS: The hemipenes are tiny (# 6.5 mm long) relative to the size of the males. The organ lacks spines of any size and is divided for nearly half its length (fig. 65). The basal stalk of the hemipenis is nude, without ornamentation; there is a pronounced swelling below the crotch, on the distal third of the asulcate side of the stalk. The hemipenial lobes are calyculate, with the calyces being confined to the separate lobes and not confluent across the hemipenial crotch. The sulcus spermaticus bifurcates proximal to the crotch; the sulcus branches extend in centrolineal orientation onto the apices of the lobes. Each sulcus branch terminates as a nude channel between dense lines of columnar papillae, which rise above the calyces on the tips of the hemipenial lobes (fig. 65C).

NATURAL HISTORY

Tropidurus bogerti is mostly found in open, rocky scrub, where the lizards are frequently seen moving about over bare rock, especially during sunny weather. They are exceedingly shy and hard to approach. Some were taken by use of a revolver with.22-caliber dustshot, but most were collected by turning sandstone rocks, where the specimens occasionally were found in association with frogs ( Leptodactylus rugosus or Tepuihyla edelcae ).

Males of this species prove to be considerably larger than females (see measurements above and fig. 64). Adult males are recognized by presence of large black patches under the thighs and around the cloaca. Scales in the black precloacal patch are irregularly covered with small tubercular scale organs (fig. 66) that seem absent in the postcloacal and thigh patches; see Harvey and Gutberlet (2000) for illustrations of other kinds of scale organs in Tropidurus .

Five mature females 67–75 mm SVL were reproductively active as shown by presence of large oviductal eggs and/or wide spent oviducts. Four females have a single large egg in either the right or left oviduct, and smaller ova in the other oviduct. The largest ovum in two individuals is round and about 8 mm or 12 mm in diameter; two other individuals each have a much larger and elongate egg in one oviduct. In one of these (fig. 64 left), the egg in the left oviduct is shelled and measures about 25 × 10 mm, whereas the largest ovum on the right side is 5 × 3.5 mm (eggs measured in situ). Such eggs that are nearly ready to be laid are very large relative to the size of the mother. It seems evident from this small sample that maturation of ova alternates between ovaries and that only one egg is laid at a time.

It follows that nests of multiple eggs must represent either repeat laying by the same female or communal nests. Two sets of three eggs each were found under sandstone rocks at Camp 3, one set already hatched and one set live. The three fresh eggs measure 25– 27 mm by 12.5–13.5 mm in preservative. Gorzula found 35 eggs under a single rock slab in May—certainly a communal nest site—the eggs ranging from fresh ones with embryos to old empty shells (Gorzula and Señaris, 1999: 155, 254).

REMARKS

Tropidurus bogerti is endemic to Auyantepui, where it may be confined to the rocky summit as assumed by Myers (1997: 1) and Gorzula and Señaris (1999: 155). However, although Roze (1958a) did not note an elevation for the holotype, it is listed as only 1100 meters in the AMNH catalog. No other specimen of T. bogerti was obtained at the 1100 m camp (Guayaraca) of the 1937–1938 expedition, nor have subsequent workers reported it from that low an elevation. Ditto marks were used in the AMNH catalog to repeat the 1100 m elevation from preceding lines for Ameiva ameiva and Tropidurus hispidus , and a simple recording error is suspected.

The key to Venezuelan tropidurid lizards in Myers and Donnelly (2001: 75) followed the generic arrangement of Frost (1992), who had synonymized Plica and Uracentron under Tropidurus . Subsequently, Frost et al. (2001) reversed this action. Current names of the Venezuelan tropidurid species are:

Plica lumaria Donnelly and Myers

Plica pansticta (Myers and Donnelly) , new combination

Plica plica (Linnaeus)

Plica umbra (Linnaeus)

Tropidurus bogerti Roze

Tropidurus hispidus (Spix)

Uracentron azureum (Linnaeus)

Uranoscodon superciliosus (Linnaeus)

Tropidurus bogerti is easily distinguished from T. hispidus by its orange-spotted color pattern and presence of tufts of spiny scales on the neck. However, three species of Plica ( lumaria , plica , and pansticta ) also are characterized by projecting tufts of elongate, spiny scales on the neck. Tropidurus bogerti is quickly distinguished from these by lack of a middorsal crest of enlarged scales and the orange-spotted color pattern.

Donnelly and Myers (1991: 32) thought that the neighboring Plica lumaria (Cerro Guaiquinima) and Tropidurus bogerti shared a color pattern of ‘‘vague pale cross-lines on a dark ground’’, although any perceived cross-lining in T. bogerti (based on variable arrangement of the pale spots) is much less evident than in P. lumaria . Nonetheless, we agree with Gorzula and Señaris (1998: 155) that ‘‘The general aspect [of T. bogerti ] is very similar to Plica lumaria ’’. Both are tepui rock dwellers and resemblances may reflect common adaptations.

Frost et al. (2001: 362) maintained Tropidurus bogerti in its own species group based on morphology (lateral gular scales imbricating posteriorly rather than laterally, and moderately depressed habitus) and color pattern (pale dots on a black ground). Combined morphology and molecular data appear to ally T. bogerti with the Tropidurus spinulosus group, whose several species occur far south of the Amazon, suggesting that T. bogerti has been long isolated in southern Venezuela ( Frost et al., 2001: 359, fig. 6).

SNAKES

FAMILY COLUBRIDAE Atractus guerreroi , new species Figures 67–69, 72

HOLOTYPE: EBRG 3403, an adult male from summit of Auyantepui at 5 ° 469N,

62 ° 329W, 2100 m elevation, Bolívar, Venezuela; collected by Ricardo Guerrero, February 6–8, 1988. (The type locality is the same as the later-established Camp 5 of the 1994 AMNH-Terramar Expedition; see map 1.) Figure 67.

ETYMOLOGY: This snake is named for our colleague and field companion Dr. Ricardo Guerrero, who collected the only known specimen several years before the AMNH–TERRAMAR Expedition.

DIAGNOSIS: An Atractus of small adult size (1 8 250 mm total length), 17 dorsal scale rows, with a dark vertebral stripe and ill-defined lateral stripes; a prominent midventral line of large black spots; supralabials 7, with labials 2–3 touching short loreal and 3–4 touching eye; maxillary teeth 6; hemipenis noncapitate, feebly bilobed, completely spiny, with a long basal naked pocket on asulcate side.

Atractus guerreroi most closely resembles its geographic neighbor Atractus steyermarki

×1.28.

Roze (Chimantá massif), which is relatively melanistic in color (dark stripes not evident or inconspicuous in life) and which has more conspicuous scale organs on some cephalic plates. Atractus guerreroi has a smaller head and shorter snout than A. steyermarki , with disproportionately smaller frontal, internasal, and loreal plates but with comparable sized parietals and prefrontals (as shown in fig. 72). See Remarks and Comparisons.

DESCRIPTION OF HOLOTYPE

The specimen is a male, undissected except for the tail; it is judged sexually mature because the hemipenial spines are well mineralized. It is slightly desiccated but otherwise in good condition; the stratum corneum is intact. See table 9 for detailed measurements.

PROPORTIONS AND SCUTELLATION: Total length 250 mm, tail length 35 mm (14.0 % of total). A small slender snake, with body slightly wider than high, rounded ventrolaterally; greatest head width 70 % of head length from snout to end of parietals and 63 % of length from snout to end of mandible; greatest body width 2.7 % of SVL. Dorsal scales smooth, lacking apical pits, in 17-17-17 rows; anal ridges present. Ventrals 166, anal plate undivided, subcaudals in 41 pairs.

Head slightly wider than neck; snout bluntly rounded in dorsal view, rounded in profile; rostral wider than high, visible from above; internasals small, slightly wider than long, half (50 %) the length of prefrontal suture; prefrontals large, longer than wide (greatest prefrontal width 80 % of greatest length); prefrontal suture asymmetrical (dextral) to the internasal suture; prefrontal suture 89 % length of frontal plate; supraoculars longer than wide; frontal wider than long, pentagonal; interparietal suture much longer than prefrontal suture and longer than frontal plate (interparietal suture 1.28 × frontal length). See figure 68 for cephalic scutellation.

Eye moderate, longer than loreal, contained 2.0 times in snout length (sagittal); eye length 1.4 times longer than its distance to lip; eye protuberant to level of lip and visible from below. Pupil subcircular (sensu Myers, 1984). Nasal divided above and below naris,

TABLE 9 Measurements a (in mm) of Holotypes of Atractus guerreroi , new species, and Atractus steyermarki Roze

its greatest length 2 times loreal length; loreal short (73 % of eye length), 1.6 times longer than its greatest height, broadly separated from internasal, entering eye; no preoculars; supralabials 7, 2nd–3rd touching loreal, 3rd– 4th touching eye; postoculars 2, the upper 1.5 times larger than lower; temporals 1 + 2, the upper one in row 2 very elongated, reaching past end of parietal.

Infralabials 7, first pair in contact behind mental, first three in contact with genials; single pair of genials, each genial 2.7 times longer than wide; four large median gulars (including two preventrals) between genials and first ventral. Head-plate tubercles tiny, inconspicuous, most concentrated on snout and chin, otherwise sparse. Some larger tubercles are discernible in a few places, particularly along the posterior edges of the prefrontals and anterolateral edges of the frontal.

COLOR PATTERN: The specimen is medium brown, with a generally poorly defined pattern of narrow brown stripes and a low lateral pale stripe (fig. 67). There is most prominently a middorsal stripe (vertebral scale row and adjacent edges of paravertebrals) extending from the parietal plates to the tail tip. And there are indications of several ill-defined lateral stripes—on scalerow 6, on the adjacent edges of rows 3–4, and on the adjacent edges of rows 2–3. The lower 1 K scale rows are white in preservative, forming a low lateral pale stripe. The lateral brown striping and the low pale stripe are anteriorly vague, becoming somewhat more pronounced posteriorly.

The white ventral surface bears a median line of prominent blackish brown markings, starting anteriorly as a small spot on the first ventral and gradually enlarging posteriorly as close-set half-moon markings that fuse to a solid stripe under the tail. The lateral edges of the ventral plates are brown (paler than the blackish midventral markings), setting off the pale lower lateral stripe.

The head is brown, with a pale area on the outer edge of each parietal anteriorly and a poorly defined pair of large pale spots posteriorly on the ends of the parietals. The brown pigmentation extends onto the upper edges of the supralabials, which otherwise are immaculate white. The anterior face of the rostral plate is dark brown, with a white lower edge. There is a pair of brown spots on the chin, each spot occupying a first infralabial and part of the adjacent genial; underside of head and neck otherwise white. A dorsad extension of white extending onto side of neck behind mouth.

MAXILLARY TEETH: Examined in situ on right side. There are six well-spaced, recurved teeth, the first three large and subequal, the last three smaller and decreasing in size, with the ultimate tooth being very small. The teeth are firmly ankylosed and there are no empty sockets.

HEMIPENIS: The noncapitate, noncalyculate hemipenis is spinose over all except the spinulate basal section. Weak bilobation, most evident in the uneverted organ, is discernible but inconspicuous after eversion. The right retracted organ was 10.5 mm long from base to the ends of very short lobes. Bilobation started at the level of subcaudal 14. The lobes were 0.3 mm long; the two slips of retractor muscle merged 1.2 mm posterior to insertion, with the major retractor originating at the level of subcaudal 27. This organ was removed, soaked in a solution of 3 % KOH for 10 hours and manually everted.

The everted hemipenis (fig. 69) is 11 mm long. The sulcus spermaticus forks more than halfway (64 %) from the base, the two branches extending in centrolineal orientation to the tips of the lobes; bilobation is negligible (, 3 % total length of organ). The distal 73 % of the organ is densely covered with spines all around, mostly medium-sized but smaller near the sulcus spermaticus; the space between the sulcate branches is sparsely covered in small spines.

The basal 27 % of the hemipenis is spinulate. There is a long basal naked pocket on the asulcate side, extending distad nearly halfway (45 %) from the base, well into the zone of spines.

REMARKS AND COMPARISONS

Atractus is the largest genus of snakes in South America, with the montane species tending to have restricted ranges. Species from other parts of Venezuela (e.g., Roze, 1966) and other parts of Guayana ( Hoogmoed, 1980; Ávila-Pires, 2005) seem not to have penetrated the uplands and highlands of the Venezuelan Guayana. A few Atractus are endemic to this region, where the genus is known primarily through the work of Janis Roze (1958b, 1961). Roze named four species— A. insipidus from Río Uraricapará on the Venezuelan-Brazilian frontier, A. steyermarki from the Chimantá massif, and A. duidensis and A. riveroi from Cerro Duida. Atractus insipidus is a spotted snake with 15 dorsal scale rows. The holotypes of the other three species, characterized by 17 scale rows, have been compared directly with the type specimen of Atractus guerreroi . Atractus riveroi is a much larger snake (486 mm total length) with transverse rows of pale spots and a dark checkered venter. Atractus duidensis is a larger brown snake (346 mm total length) without markings dorsally; ventral surfaces are dark, with suffusions of brown extending transversely across the ventral plates. Only Atractus steyermarki bears resemblance to A. guerreroi .

NOTES ON ATRACTUS STEYERMARKI : Roze (1958b) named Atractus steyermarki based on two specimens from elevations of 1430 m and 2160 on Chimantá-tepui. Later, Roze (1961: 116; 1966: 87) mentioned that the species had been rediscovered by Steyermark ‘‘en la región de El Dorado’’. The lowland town of El Dorado, about 150 km NE Chimantá, lies at an elevation of 130 m, but elevations above 1200 m occur well to the south along the ‘‘El Dorado– Santa Elena de Uairén road’’ (see the regional description in Duellman, 1997). Roze took the locality at face value and assumed a broad distribution ‘‘en la Gran Sabana en general’’, but he provided no descriptive data for the specimen. The El Dorado snake, cited by Roze (1966: 87) as an unnumbered MBUCV specimen (Museo de Biología de la Universidad Central de Venezuela, Caracas), apparently has not been seen by other Venezuelan workers (e.g., Lancini and Kornacker, 1989; Gorzula and Señaris, 1999).

Gorzula (1992: 272) reported an additional specimen of Atractus steyermarki from Chimantá Camp VIII (5 2600 m fide Huber, 1992: 31). Gorzula and Señaris (1999: 161– 162) also provided data for this specimen and mentioned the El Dorado record as unverified, concluding that ‘‘the available evidence indicates that this species is endemic to the highland summits of the Chimantá Massif’’. The known elevation range for Atractus steyermarki on Chimantá is 1430–2600 m.

The holotype of Atractus steyermarki (figs. 70, 71) has been examined; see table 9 for detailed measurements. The specimen is in fair condition and pliable, although somewhat desiccated; it evidently has faded sufficiently in preservative to reveal a basic striped pattern (see below). There are six maxillary teeth, which appear much as described for A. guerreroi above, and the hemipenis also is similar to that of A. guerreroi . Unfortunately, both hemipenes are damaged and can not be easily illustrated. The left organ had been opened irregularly from the medial side and is torn. The proximal section of the right organ is mutilated, although the distal part (opened midventrally) is in good condition; the right organ extends to subcaudal 13. The hemipenes are spinose and very feebly bilobed; the attachment slips of retractor muscle are correspondingly very short.

The male holotype (‘‘270’’ [257] mm total length) and female paratype (420 mm) of Atractus steyermarki were described by Roze (1958b) as patternless dark gray, turning paler on the lower 1–2 scale rows. The midventral dark spots in the male cover most of the ventrals in the larger, more melanistic female. Gorzula (1992: 272) described the living colors of his intermediate-sized specimen (a male, 373 mm total length) as dark brown above, yellow below, with dark brown pigment midventrally.

The aforesaid descriptions document Atractus steyermarki as having melanistic tendencies that evidently cause the dorsum to appear essentially unicolor in life. Nonetheless, after a half century of preservation, the holotype has faded sufficiently for a striped pattern to appear (especially when immersed in alcohol, see fig. 70). Three prominent dark brown stripes are now visible—a middorsal stripe confined to the scales of the vertebral row, a high lateral stripe centered on row 6 and overlapping onto adjacent edges of rows 5 and 7, and a lateral stripe on row 3 and adjacent edge of row 2. Scales in rows 1 and 2 are mostly pale gray (with brown pigment at the scale base), appearing as a low pale lateral stripe set off by the brown stripe above and black ventral tips below. The midventral marking starts as small spots on ventral 13, enlarging posteriorly as black half-moons that nearly fuse to form a median stripe.

COMPARISON OF ATRACTUS STEYERMARKI AND A. GUERREROI : As mentioned above, the maxillary teeth and hemipenes of Atractus steyermarki appear to be essentially the same as in A. guerreroi . The midventral dark pattern also is essentially the same in Atractus guerreroi and A. steyermarki . The dark dorsal stripes and ill-defined pale ‘‘stripe’’ on the first two scale rows of A. guerreroi are similar to the above-described pattern on the faded holotype of A. steyermarki , but there are differences.

The middorsal stripe of A. guerreroi is microscopically less dense and overlaps onto the paravertebral rows. Both have a stripe on row 6, but in steyermarki it is better defined and wider, overlapping onto the adjacent rows. Atractus guerreroi lacks the vivid lateral dark stripe (row 3 and edge of row 2) of steyermarki , having instead two very illdefined lateral stripes or dark suffusions along the adjacent edges of rows 3–4 and rows 2–3. These differences in lateral stripes might be diagnostic, but, at this time, there is no way of assessing variation within populations. In A. steyermarki , the conspicuous stripes on the faded holotype seem likely to reflect a juvenile pattern that becomes ontogenetically obscured by melanism.

The holotypes of both Atractus guerreroi and A. steyermarki have the very minute, inconspicuous cephalic sensory tubercles that are widespread among colubrids. Both also have a larger kind of tubercle, which, in A. guerreroi , is discernible especially along the posterior edges of the prefrontals and anterolateral edges of the frontal. These organs are blackish and more numerous—or at least much more conspicuous—in A. steyermarki , where they cause the snake to look as if some of its head plates had been ‘‘riveted on’’. 33 in

33 We borrow this apt phrase from unpublished notes by Frances J. Irish, who observed the presence of these usually overlooked structures in the types of Atractus steyermarki and in various species thought to be allied with A. collaris . Irish judged the structures as possibly homologous with apical pits on the body scales of some species; they seem to vary in appearance from pitlike to tuberclelike. They are easily seen under the dissecting microscope in Atractus steyermarki because of associated black pigmentation. They can be difficult to see in other species unless scale surfaces are blotted dry and the angle of light carefully adjusted.

A. steyermarki , the structures are prominent along the posterior edges of the rostral, internasals, and prefrontals, somewhat less prominent along the lateral edges of the frontal, and conspicuously present laterally on the top edge of the loreal, posterodorsal edge of the nasal, and on the upper parts of supralabials 2–3 (fig. 71).

The male holotypes of Atractus steyermarki and A. guerreroi are of comparable size, the former only 4 mm longer in SVL (3 mm longer in tail length), but A. guerreroi appears to have a disproportionately smaller head, as shown by measurements (table 9) and as readily visualized in side-by-side comparison (fig. 72). The parietals and prefrontals appear to be of comparable size in A. guerreroi and A. steyermarki , but the internasals and frontal are disproportionately smaller in A. guerreroi . Atractus guerreroi has a relatively shorter snout, as also reflected in a shorter loreal, which is 64 % of eye length in guerreroi vs. 92 % in steyermarki . Both have 7 supralabials, but there is a slight difference in the configuration. In A. guerreroi , the first supralabial is far enough removed from the loreal that contact might seem unlikely in intraspecific variation, whereas in A. steyermarki the first supralabial is in point contact with the loreal on the left side and nearly in contact on the right side.

In summary, Atractus guerreroi most closely resembles A. steyermarki . The limited data suggest to us that the evolutionary course of A. steyermarki has been toward melanism and that of A. guerreroi toward reduced head size—hypotheses needing larg- er sample sizes.