Ontogeny of diving and feeding behavior in juvenile seaturtles: Leatherback Seaturtles (Dermochelys coriacea L) and Green Seaturtles (Chelonia mydas L) in the Florida Current

We compared activity, diving behavior and response to prey by Dermochelys coriacea and Chelonia mydas during their first 8-10 weeks of development.We reared juveniles in the laboratory and, at two-week intervals, released them in the ocean for a brief trial. Each turtle towed a device used to measure its dive profile. All turtles swam throughout their trials, but D. coriacea swam more slowly than C. mydas. Dermochelys coriacea dives had V-shaped profiles and older turtles made longer and deeper (up to 18 m) dives than younger turtles. Chelonia mydas dives were usually shallow (<6 m) and consisted of three (V, S, and U) profiles. Older C. mydas made dives that were longer but only slightly deeper than those of younger turtles. Dermochelys coriacea fed throughout the water column exclusively on gelatinous prey Aurelia, ctenophores, and unidentified gelatinous eggs. Chelonia mydas fed near the surface on floating Thalassia and Sargassum or at shallow depths on ctenophores and unidentified gelatinous eggs but ignored large jellyfish (Aurelia). Thus, early in development the two species overlap in foraging area and in diet. However as D. coriacea grow they dive deeper where prey assemblages probably differ from those in shallow water where C. mydas feed. These distinct behavioral trajectories probably cause the niches of D. coriacea and C. mydas to separate spatially very early in their development.

journal of Herpetology, Vol. 38, No. 1, pp. 36-43, 2004 Copyright 2004 Society for the Study of Amphibians and Reptiles Ontogeny of Diving and Feeding Behavior in Juvenile Seaturtles: Leatherback Seaturtles (Dermochelys coriacea L) and Green Seaturtles (Chelonia my das L) in the Florida Current Michael Salmon,1'2 T. Todd Jones,1 and Kenneth W. Horch3 lDepartment of Biological Sciences, Florida Atlantic University, 777 Glades Road, Box 3091, Boca Raton, Florida 33431-0991, USA JDepartment of Bioengineering, University of Utah, Salt Lake City, Utah 84112, USA We compared activity, diving behavior and response to prey by Dermochelys coriacea and Chelonia mydas during their first 8-10 weeks of development. We reared juveniles in the laboratory and, at two-week intervals, released them in the ocean for a brief trial. Each turtle towed a device used to measure its dive profile. All turtles swam throughout their trials, but D. coriacea swam more slowly than C. mydas. Dermochelys coriacea dives had V-shaped profiles and older turtles made longer and deeper (up to 18 m) dives than younger turtles. Chelonia mydas dives were usually shallow (<6 m) and consisted of three (V, S, and U) profiles. Older C. mydas made dives that were longer but only slightly deeper than those of younger turtles. Dermochelys coriacea fed throughout the water column exclusively on gelatinous prey Aurelia, ctenophores, and unidentified gelatinous eggs. Chelonia mydas fed near the surface on floating Thalassia and Sargassum or at shallow depths on ctenophores and unidentified gelatinous eggs but ignored large jellyfish (Aurelia). Thus, early in development the two species overlap in foraging area and in diet. However as D. coriacea grow they dive deeper where prey assemblages probably differ from those in shallow water where C. mydas feed. These distinct behavioral trajectories probably cause the niches of D. coriacea and C. mydas to separate spatially very early in their development The beaches of southeastern Florida serve as rookeries for populations of Green Seaturtle (Chelonia mydas L.), Leatherback Seaturtle (Der­mochelys coriacea V.) and Loggerhead Seaturtles (Caretta caretta L.; Meylan et al., 1995). After emerging from nests, their hatchlings crawl to the surf zone, then migrate offshore during a 24-36 h "swimming frenzy" (Wyneken and Salmon, 1992). Hatchlings eventually encounter the Flor­ida Current (western portion of the Gulf Stream) that probably carries them to open water "nursery" habitats (Witham, 1980; Musick and Limpus, 1997). Except for C. caretta, the geo­graphic location of marine turtle nursery habitat in the pelagic is unknown, as is how and where foraging occurs in the water column. Both sources of information are essential for defining the ecological niche of D. coriacea and C. mydas (Brown, 1995). Juvenile D. coriacea (<100 cm in curved carapace length [CCL]) have been seen at low latitudes (6-14°) in the open ocean where water temperatures are relatively high (26-36°C). As they grow sightings shift to higher latitudes where temperatures are variable and often cooler (Eckert, 1999). Swimming D. coriacea likely stay 2 Corresponding Author. E-mail: [e-mail] in open water where there are no obstructions because in captivity they do not recognize and have difficulty escaping from physical barriers (Witham, 1977). Posthatchling C. mydas are so rarely observed in the ocean that their nursery habitats are unknown. Some sightings have been near Sar­gassum driftlines (Caldwell, 1969; Carr, 1986b; Carr and Meylan, 1980) but the counter-shading of young C. mydas indicates they probably forage in open water (Musick and Limpus, 1997). In laboratory studies, C. mydas avoid contact with flotsom (Mellgren et al., 1994). Witherington (2002) found no posthatchling C. mydas where C. caretta of that age were abundant (drift lines in the Florida Current), suggesting the two species occupy different habitats. To our knowledge, the behavior of posthatch­ling D. coriacea and C. mydas has not been observed in deep, open ocean habitats. We reared hatchlings of both species in the laboratory to gain insights into how the two species might differ in habitat use. At intervals of approxi­mately two weeks, turtles were transported several kilometers offshore to the Florida Current to observe their diving and feeding behavior. Our goals were to determine (1) how the two species differed in their use of space while swimming, diving, and feeding, and (2) how their behavior changed as they increased in size and age.ONTOGENY OF BEHAVIOR IN SEATURTLES 37 Maintenance.-We reared 34 hatchlings (1-8 per nest from 11 nests) of D. coriacea, using previously described methods (Jones et al., 2000). Hatchlings were obtained from nests located in Palm Beach County, Florida, between July and August 1998 and 1999. We individually marked hatchlings on the carapace with distinct dot patterns of white correction fluid ("liquid pa­per"). Hatchlings were housed in groups of four with circular plastic pools (1.5 m wide X 30 cm deep). Pools were filled with 103 liters of filtered and sterilized seawater, partially changed every three days. Water temperature was maintained between 23° and 27°C. Turtles were fed once daily to satiation using a gelatin-based artificial diet (Jones et al., 2000). We captured 33 C. mydas (4-13 per nest) as they emerged between August and September from five nests in Palm Beach County. Turtles were placed in a large, circular (2.5 m wide X 30 cm deep) plastic pool, continuously supplied with fresh, sand-filtered seawater. We marked each hatchling with a small (7.0 mm2) uniquely colored and numbered plastic strip fixed to its carapace with a drop of cyanoacrylic cement. We fed juveniles (defined as posthatchlings that had commenced feeding) raw shrimp once daily to satiation, and maintained the turtles on a 14:10 L:D photoperiod at 22-26°C. Dermochelys coriacea used in trials were (age in weeks/number of turtles): 2/5, 4/12, 6/7, 8/6, and 10/4. At two weeks of age, juveniles were (mean ± SD) 59.6 ± 9.45 g in mass and 72.5 ± 5.0 mm in SCL. At 10 weeks of age, the turtles more than doubled in mass and increased in SCL by ~ 30%. Growth curves for this species are presented elsewhere (Jones et al., 2000). Chelonia mydas used in trials were 2/9, 4/8, 6/8, and 8/8. Turtles increased from 34.8 ± 6.6 g in mass and 62 ± 3.5 mm in SCL at two weeks of age to 70.2 ± 24.4 g and 79.1 ± 11.4 mm in SCL at eight weeks of age. Field Trials.-Between late August and early November, we transported 2-4 juveniles from two or more nests to sites 4-6 km offshore and within the Florida Current (depth: 100-200 m). Turtles were not fed for 24 h prior to transport. Sites were adjacent to Boynton Beach, Palm Beach County, Florida (26°30'N lat., 80°00'W long.). On the boat the turtles were housed in a 182-liter live well, filled with continuously circulating seawater. Total transport time was <1.5 h. Releases took place during the midmorning or early afternoon. Dermochelys coriacea were given a 30-min trial. Chelonia mydas were given 35-min trials because they sometimes made a single long, deep dive immediately after release. Thereafter, Materials and Methods their dives were relatively shallow and short. Therefore, we gave them 5 min to acclimate to their oceanic surroundings. At the conclusion of their trial (or their last dive, if it was initiated just before the trial ended), juveniles were released (after removing plastic tags or white markings). During trials, two observers wearing masks and snorkels followed 4-6 m behind the turtles. Both observers remained at the surface. One held a spool containing 70 m of 7-kg test line, which was attached to our depth-recording device towed by the turtle. Sufficient line was released while the turtle was diving (swam downward to a depth >0.5 m) so that its descent was un­impeded. Line was retrieved when the turtle surfaced. The second observer used hand signals to indicate the beginning and end of each dive to a third observer, driving the boat about 10 m behind both swimmers. Water clarity was suffi­cient for swimmers to see diving juveniles during all but their deepest dives. Dive Measurement.-We measured dive pro­files using two devices. In 1998, 14 D. coriacea towed an air-filled aluminum cigar tube 150 mm long X 15 mm in diameter. We glued a 1.0 m length of 2.3-kg test monofilament to the front of the tube and attached the other end to the turtle's carapace with a small Velcro™ patch. This line was designed to break if a predator took the turtle. The cigar tube was rendered neutrally buoyant by the addition of internal weights and balanced so that the tube floated horizontally in the water column. Dive profiles were measured by pointing a sonar unit (Furuno FE-4200, Hyogo, Japan) at the cigar tube during the dive. Records were stored on a chart recorder housed on the boat. During the 1999 field season, 20 D. coriacea towed a miniature time-depth recording (TDR) tag (Lotek LTD 100, Newfoundland, Canada; 57 mm long X 18 mm in diameter; mass under water, 1 g). The tag was attached to the turtle by a 1.0-m length of fine monofilament, ending in a velcro patch that adhered to its carapace (Fig. 1). In 2000, we used the tag to measure C. mydas dives. It was attached to the turtle by a 1.0-m length of fine monofilament line that encircled its body behind the foreflippers. We made the tag neutrally buoyant for D. coriacea (whose consecutive dives were usually separated by several min at the surface) and slightly positively buoyant for C. mydas. Positive buoyancy enabled the tag to reach the ocean surface between consecutive C. mydas dives, which often were initiated after only a short (1­2 sec) pause to breathe. The TDR recorded pressure and time at 10-sec (D. coriacea) or 2-sec (C. mydas) intervals. Records were downloaded and stored on a computer.38 M. SALMON ETAL. Behavioral observations were documented as field notes and by underwater photography. Drag imposed by the tag, tether, and cigar tube probably decreased swimming speed and may have affected dive depth. We balanced these disadvantages against the following advantages. First, tethering eliminated the possibility that a turtle might escape during a trial by diving beyond our visual range. Second, a TDR was the least intrusive method for measuring dives. The alternative (following the turtle using SCUBA and completing measurements by instruments we carried) increased the possibility that the observer might affect the turtle's behavior. Third, our purpose was to determine how the two species differed in diving, swimming, and associated behavior, not to measure maximum swimming speed or dive depth. Data Analyses.-Dive parameters (descent speed; duration; depth; profiles) were measured from the TDR or chart records. Descent speed (cm/sec) was based upon the actual (for turtles diving once) or average (for turtles diving more than once) time to descend to depth. Dive profiles were classified using the criteria of Hochscheid et al. (1999). Statview© (Abacus Concepts, Berkeley, CA) was used to compute correlation coefficients for dive parameters (duration vs. depth). Mean dive depth, dive duration and descent speed was calculated for each turtle. An ANCOVA was then used to determine whether the two species differed in these attributes as a function of age, with mass as the covariate. Mann-Whitney, Kruskall-Wallis, and Chi-square tests (corrected for continuity; Siegel and Castellan, 1988) were used to specify more precisely how dives changed with age. Results Swimming Behavior.-Juveniles of both species swam and made dives by "powerstroking" (synchronous movements of the foreflippers; Wyneken, 1997). During trials, D. coriacea swam slowly near the surface with little change in direction. Diving was accomplished with no obvious change in stroke rate or swimming speed. Chelonia mydas also swam near the surface but more rapidly, often changing speed and direction. Diving was initiated by a brief but abrupt increase in stroke rate and speed, fol­lowed by continued descent to depth at a slower speed. Diving Frequency.-Twenty-one of 34 D. cor­iacea (62%) made dives (mean ± SD for the diving turtles = 4.10 ± 2.53 dives/trial, range = 1-9; N = 86 dives). Diving frequency did not change with age (nine of 17 turtles < 4 weeks vs. 12 of 17 turtles > 4 weeks of age; n.s. by a x2'test). All c // / > * Fig. 1. Green turtles (A, B) and a leatherback (C) in the Florida Current. (A) Turtle towing a TDR. Camera angle makes the tag appears large relative to the turtle. (B) Chelonia mydas feeding on gelatinous eggs near the surface (note expanded gular region). (C) Dermochelys coriacea feeding on Aurelia.ONTOGENY OF BEHAVIOR IN SEATURTLES 39 Fig. 2. Dive profiles. Top and middle/ W- and V- dives of two Dermochelys coriacea. Bottom: S-, V-, and U-dives of a green turtle. Consecutive readings (filled diamonds) are at 10-sec {Dermochelys coriacea) and 2-sec (Cheknia mydas) intervals. green turtles made dives during their trials (9.1 ± 5.46; range = 1-25; N = 299). Diving frequency did not change with age (148 dives by 17 turtles < 4 weeks vs. 151 dives by 16 turtles > 4 weeks; n.s. by a %~-test). The two species differed significantly in the number of dives made by diving turtles (86 dives by 21 D. coriacea-, 299 dives by 33 C. mydas; %2 = 15.14, P < 0.001,1 df). Dive Profiles.-Profiles were obtained for 71 of 86 D. coriacea dives. Most (N = 62, or 87 %) were V-dives (Fig. 2); the remainder were W-dives that included small changes in vertical position once turtles reached depth. In five W-dives, the turtles were feeding at depth on jellyfish. Chelonia mydas showed three dive profiles: V (N = 141), U (N = 119), and S (N = 20; Fig. 2). Nineteen others could not be classified. S-dives were 2.5-5.5 m in depth, and 20-200 sec in duration while V- and U-dives were 0.5-9.3 m in depth and 15-260 sec in duration. However, over 90% of these dives were shallower (<5.0 m) and shorter (<180 sec). V-dives were significantly shallower (Mann-Whitney Z = -4.35, P < 0.001) and shorter (Z = -7.52, P < 0.001) than U-dives. Two C. mydas made especially deep V-dives during the acclimation period and two after their trial ended (to evade recapture). These were the longest (561 sec by an eight-week-old turtle) and deepest (18.7 m by a four-week-old turtle) dives we observed in this species. Comparisons in Descent Speed, Duration, and Dive Depth.-The two species differed signifi­cantly in descent speed (Table 1). Average descent speed (mean ± SD) for 21 D. coriacea was 7.08 ± 3.20 cm/sec (range: 3.4-16.2 cm/sec). For 32 C. mydas, it was 21.26 ± 9.66 cm/sec (range: 8.2-40.8 cm/sec). Dive duration was positively and significantly correlated with dive depth in both D. coriacea (N = 86, r = 0.82, P < 0.01) and C. mydas (N = 299, r = 0.69, P < 0.01; Fig. 3). Most dives by both species were shallow (<6 m) but in D. coriacea many were deeper (up to 17.1 m). Chelonia mydas did not dive deeper than 9.3 m (Fig. 3). Dive duration (Table 2) and dive depth (Table 3) differed significantly between the species. In D. coriacea, dives were on average deeper (5.29 ± 3.37 m) and longer (128.14 ± 84.38 sec) than those of C. mydas (2.47 ± 1.16 m, 82.21 ± 27.60 sec). Diving as a Function of Age.-Descent speed (Table 1), dive duration (Table 2), and dive depth did not differ significantly with age (Table 3). However, these analyses were based upon means of each behavior for individual turtles. The raw data (Fig. 3) suggested that at least for D. coriacea, older turtles most often made the deepest and longest dives. Because in both species most dives were relatively short (<200 sec) and shallow (<6 m), mean values may have masked these contrasts. The data were reexamined by comparing the distribution of the deepest and longest dive made by each turtle, as a function of its age. In D. coriacea, the deepest dive made by each younger (2-6 weeks) turtle ranged between 1.2 and 15.2 m (N = 13 turtles, mean = 5.75 m), whereas the deepest dive made by each older (7-10 weeks) turtle ranged between 6.3 and 17.1 m (N = 8, mean = 10.65 m). Older turtles made signifi­cantly deeper dives (Wilcoxon Z = -2.68, P = 0.003). The longest dive made by each younger D. coriacea ranged from 15-360 sec (N = 13,40 M. SALMON ET AL. Table 1. ANCOVA summary table for determining the effects of species, age, and their interactions on diving descent speed of Chelonia mydas and Dermo­chelys coriacea, 2-8 weeks of age. Source of variation MS df F P Covariate (mass) 218.50 1 3.74 0.061 Species 1433.98 1 20.70 0.0001 Age 178.39 3 1.25 0.307 Species X age 39.53 3 0.1 0.189 Table 2. ANCOVA summary table for determining the effects of species, age, and their interaction on dive duration of Dermochelys coriacea and Chelonia mydas, 2-8 weeks of age. Source of variation MS df F P Covariate (mass) 9334 1 2.31 0.136 Species 29077 1 7.96 0.007 Age 28479 3 2.52 0.072 Species X age 14678 3 1.67 0.189 Fig. 3. Dive depth and duration shown by Der­mochelys coriacea (above, N = 45 dives by younger and 41 dives by older turtles) and by Chelonia mydas (lower, N = 148 dives by younger and 151 dives by older turtles). Table 3. ANCOVA summary table for determining the effects of species, age, and their interaction on dive depth of Dermochelys coriacea and Chelonia mydas, 2-8 weeks of age. mean = 119.23 sec) whereas among older turtles the range was 170-360 sec (N = 8, mean = 256.5 sec). Older turtles made longer dives (Wilcoxon Z = -3.15, P = 0.001). In C. mydas, the same trends were apparent although absolute differences between the age groups were less extreme. The deepest dive made by each younger (2-4 weeks) turtle ranged between 1.3 and 9.1 m (N = 17 turtles, mean = 5.07 m), whereas among older (6-8 weeks) turtles, it ranged between 1.1 and 9.3 m (N = 16, mean = 5.54 m). Older turtles made significantly deeper dives (Wilcoxon Z = -1.80, P = 0.04). The longest dive made by each younger C. mydas ranged from 40-204 sec (N = 17, mean = 102 sec), whereas among older turtles, the range was 68-258 sec (N = 16, mean = 149.6 sec). Older turtles made significantly longer dives (Wilcoxon Z = -2.72, P = 0.003). Although V-dives were the dominant profile among D. coriacea regardless of age, dive profiles changed with age in C. mydas. Younger (2-4 weeks; 148 dives observed) juveniles made more V- (N = 99) than U- (N = 40) dives, whereas older (6-8 weeks; 151 dives observed) juveniles made fewer V- (N = 42) than U- (N = 79) dives (%2= 33.3, P < 0.001,1 df). Feeding Behavior.-Nine D. coriacea fed during trials. Two ate ctenophores (Ocyropsis or Mne- Source of variation MS df F P Covariate (mass) 15.888 1 1.32 0.258 Species 54.069 1 14.92 0.0001 Age 29.474 3 1.33 0.278 Species X age 22.604 3 1.43 0.248 niopsis) and one ate gelatinous (probably mollus- can) eggs. Six turtles fed on moon jellyfish (Aurelia; Fig. 1). Feeding was observed at depths between 0.5 and 14 m. Seven C. mydas fed during their trials, all <2 m below the ocean surface (Fig. 1). Two turtles nibbled on floating Sargassum and Thalassia leaves suspended just below the surface. One turtle consumed a ctenophore and four others fed on gelatinous eggs. Chelonia mydas ignored Aurelia, even when jellyfish were abundant. Discussion Constraints.-All juveniles were reared in shallow tanks and had no opportunity to dive before they were released in the ocean, which might have modified their performance. In addition, differences in how long turtles were held in captivity might affect their performance at different ages. However, all turtles swam normally, most made dives, and some consumed natural prey. We saw no behavior that appeared "abnormal." Differences between species inONTOGENY OF BEHAVIOR IN SEATURTLES 41 behavior seemed appropriate given our current understanding of their ecological specializations. These results suggest that although turtles were reared under artificial conditions, unique fea­tures of their behavior persisted. Activity.-Previous studies on D. coriacea and C. mydas have revealed correlations between activity level and locomotion on the one hand and O2 consumption on the other. Wyneken (1997) classified D. coriacea as "marathon" and C. mydas as "sprinter" strategists with regard to their locomotion and energetics. During their frenzy period, D. coriacea hatchlings swim at relatively slow (~25 cm/sec), but constant speeds, consume little oxygen, and show small changes in O2 consumption between rest and activity ("narrow" aerobic scope; Wyneken and Salmon, 1992; Wyneken, 1997). Chelonia mydas during their frenzy period swim faster (~ 44 cm/ sec), consume 2x more oxygen than D. coriacea and show a larger change in O2 consumption between rest and activity ("wide" aerobic scope). Postfrenzy C. mydas consumed more oxygen than postfrenzy D. coriacea (Wyneken, 1997). Our observations were both qualitatively and quantitatively consistent with these assessments. Juveniles of both species swam throughout trials, but D. coriacea swimming and diving speeds were slower than those shown by C. mydas. Many D. coriacea did not dive during their trials, whereas all of the C. mydas did. Diving green turtles also averaged almost twice as many dives/trial than did diving D. coriacea. Most D. coriacea dives were short (< 100 sec in duration) and completed over a small portion of the trial. Few dives made by either species resulted in prey capture, although some prey consumed by both species were always present. Thus, differences in diving frequency between species probably were not caused by variation in prey abundance. Dive Profiles.-V-dive profiles of juvenile D. coriacea were similar to those of adult C. mydas diving in deep water (Eckert et al., 1986, 1989). The few W-dives we witnessed have not been seen in adult turtles. In more than half (five of nine) of these dives, juveniles were feeding. We suggest that W-dives occur when juveniles either find food, or while they assess prey abundance at depth (as occurs in other diving vertebrates; Thompson and Fedak, 2001). V-dives were rarely associated with feeding, and so their function remains unclear. Juvenile C. mydas performed V-dives that, in contrast to D. coriacea, were typically shallow and brief. They also showed S- and U- dives that were never observed in juvenile D. coriacea, but occur among internesting C. mydas (Hochscheid et al., 1999; Hays et al., 2000) in shallow water. Dermochelys coriacea also make U-dives under the same conditions (Eckert, et al., 1996). Whether profiles of juvenile C. mydas are functionally similar to those of adults remains unknown. Juveniles showed U- and S-dives even in deep water, where there was no depth barrier. Age, Depth, and Duration.-The shallow dives of juvenile turtles might reflect physiological constraints to deep diving. Alternatively, they might result from a behavioral preference for shallow dives that are well within physiological limits. Larger marine turtles make deeper (leath- erbacks; Eckert et al., 1986) and longer (hawks- bills; van Dam and Diez, 1997) dives than smaller conspecifics. In general, small body size should limit diving depth and duration because (1) the volume of tissue to store oxygen is lower, and (2) mass-specific metabolic rates of smaller ani­mals are higher (Schmidt-Nielsen, 1997). Thus, we expected that larger and older D. coriacea and C. mydas would show longer and deeper dives than younger turtles; this sprediction was ful­filled. In both species, older turtles most often made the deepest and longest dives. This tendency, however, was more obvious among D. coriacea (whose dives exceeded 17 m) than among C. mydas (whose dives did not exceed 9 m; Fig. 3). Most D. coriacea dive profiles were V-shaped, and therefore longer dives were accompanied by a proportional increase in dive depth. In older C. mydas, however, most dives had U-shaped profiles so that longer dives were accompanied by only a modest increase in depth. Although juvenile size must ultimately limit dive depth and duration in both species, the contrasts we witnessed are best explained by behavioral preferences. Chelonia mydas, when sufficiently motivated, were capable of dives that were comparable in depth and duration to the deepest dives made by leatherbacks. In the absence of that motivation (i.e., during "routine" diving), C. mydas chose to remain closer to the ocean surface. Behavioral Ecology of Diving.-Our results suggest that, during their first weeks in the pelagic, D. coriacea diving capacities increase rapidly with growth. During this time, they feed opportunistically on prey located between the surface and the dive depths that they can achieve, which probably are within ~20 m of the surface. "Outgrowing" these constraints may select for faster growth rates in D. coriacea than in the hard-shelled marine turtles (Rhodin, 1985). We probably witnessed the onset of that process, one which ultimately leads to adult turtles feeding over a broad range of depths (near the surface [Eisenberg and Frazier, 1983; Limpus, 1984; Grant and Ferrell, 1993; Grant et al., 1996] and in the deep-scattering layer [Eckert et al.,42 M. SALMON ET AL. 1986, 1989]) and latitudes (Bleakney, 1965; Pritchard, 1976; Brongersma, 1972; Hodge, 1979). For C. mydas, the trend was clearly different. As turtles aged, dive duration increased faster than dive depth. Pelagic-stage C. mydas probably search for and forage upon prey found relatively close (<5 m) to the ocean surface. Thus, even at an early age D. coriacea and C. mydas differed in where they foraged and in the prey that they consumed. Dermochelys coriacea fed exclusively on gelatinous prey, whereas C. mydas foraged on floating or suspended algae, as well as smaller animal prey at or near the surface. Similar results have been reported elsewhere (Hughes, 1974; Frick, 1976), suggesting that young C. mydas are omnivores with a tendency toward carnivory, whereas young D. coriacea are gelatinivores (Bjorndal, 1997). Differences in prey category, prey size, and foraging depth appear to be primary niche axes separating these pelagic turtles. Acknowledgments.-This study was supported by the National Save the Sea Turtle Foundation, and by personal funds. K. Rusenko (Gumbo Limbo Nature Complex), L. Wood, and C. Johnson (Marinelife Center of Juno Beach) supplied us with green turtle hatchlings. B. Kenyon and S. Weege drove our boat. A. Nash and D. Owen provided statistical advice. Critical readings by K. J. Lohmann, J. Wyneken, and three anonymous reviewers improved the man­uscript. This study was approved by the Florida Atlantic University Animal Care Committee, and was conducted under Florida FWCC permit TP-073. Literature Cited Bjorndal, K. A. 1997. Foraging ecology and nutrition of sea turtles. In P. L. Lutz and J. A. Musick (eds.), The Biology of Sea Turtles, pp. 199-232. CRC Press, Boca Raton, FL. Bleakkey, J. S. 1965. Reports of marine turtles from New England and Eastern Canada. Canadian Field Naturalist 79:120-128. Brongersma, L. D. 1972. European Atlantic turtles. Zoologische Verhandelingen 121:1-318. Brown, J. H. 1995. Macroecology. Univ. of Chicago Press, Chicago. Caldwell, D. K. 1969. Hatchling Green Sea Turtles, Chelonia mydas, at sea in northeastern Pacific Ocean. Bulletin of the Southern California Academy of Sciences 68:113-114. Carr, A. 1986. New perspectives on the pelagic stage of sea turtle development. NOAA Technical Mem­orandum NMFS-SEFC-190. Washington, D.C. Carr, A., and A. Meylan. 1980. Evidence of passive migration of green turtle hatchlings in Sargassum. Copeia 1980:366-368. Eckert, S. A. 1999. Global distribution of juvenile Leatherback Sea Turtles. Hubbs Sea World Research Institute Technical Report 99-294, 14 pp. La Jolla, California. Eckert, S. A., D. W. Nellis, K. L. Eckert, akd G. L. Kooymak. 1986. Diving patterns of two Leather- back Sea Turtles (Dermochelys coriacea) during intemesting intervals at Sandy Point, St. Croix, U. S. Virgin Island. Herpetologica 42:381-388. Eckert, S. A., K. L. Eckert, P. Paganis, akd G. L. Kooymak. 1989. Diving and foraging behavior of Leatherback Sea Turtles (Dermochelys coriacea). Canadian Journal of Zoology 67:2834-2840. Eckert, S. A., H. C. Liew, K. L. Eckert, akd E. H. Chak. 1996. Shallow water diving by leatherback turtles in the South China Sea. Chelonian Conservation and Biology 2:237-243. Eisekburc, J. F., akd J. Frazier. 1983. A Leatherback Turtle (Dermochelys coriacea) feeding in the wild. Journal of Herpetology 17:81-82. Frick, J. 1976. Orientation and behaviour of hatchling Green Turtles (Chelonia mydas) in the sea. Animal Behaviour 24:849-857. Grakt, G. S., akd D. Ferrel. 1993. Leatherback Turtle, Dermochelys coriacea (Reptilia: Dermochelidae): notes on near-shore feeding behavior and associa­tion with cobia. Brimleyana 19:77-81. Grakt, G. S., H. Malpass, akd J. Beasley. 1996. Correlation of Leatherback Turtle and jellyfish occurrence. Herpetological Reviews 27:123-125. Hays, G. C., C. R. Adams, A. C. Broderick, B. J. Godley, D. J. Lucas, J. D. Metcalfe, akd A. A. Prior. 2000. The diving behaviour of Green Turtles at Ascension Island. Animal Behaviour 59:577-586. Hochscheid, S., B. J. Godley, A. C. Broderick, akd R. P. Wilsok. 1999. Reptilian diving: highly variable dive patterns in the Green Turtle Chelonia mydas. Marine Ecology Progress Series 185:101-112. Hodce, R. P. 1979. Geographic distribution: Dermo­chelys coriacea schlegeli. Herpetological Reviews 10: 102^ Hughes, G. R. 1974. The sea turtles of south-east Africa. II. The biology of the Tongaland Loggerhead Turtle Caretta caretta L. with comments on the Leatherback Turtle Dermochelys coriacea L. and the Green Turtle Chelonia mydas L. in the study region. South African Association of Marine Biological Research, Ocean Research Institute 36:1-96. Jokes, T. T., M. Salmok, J. Wykekek, akd C. Johksok. 2000. Rearing Leatherback hatchlings: protocols, growth and survival. Marine Turtle Newsletter 90:3-6. Limpus, C. J. 1984. A benthic record from neritic waters for the leathery turtle (Dermochelys coriacea). Copeia 1984:552. Mellcrek, R. L., M. A. Mann, M. E. Bushonc, S. R. Harkins, and V. K. Krukke. 1994. Habitat selection in three species of captive sea turtle hatchlings. In K. A. Bjorndal, A. B. Bolten, D. A. Johnson, and P. J. Eliazar (Comp.), Proceedings of the 14th Annual Symposium on Sea Turtle Biology and Conserva­tion. NOAA Technical Memorandum NMFS- SEFSC-351. Meylak, A., B. Schroeder, akd A. Mosier. 1995. Sea turtle nesting activity in the state of Florida 1979­1992. Florida Marine Research Institute Publication 52:1-51.ONTOGENY OF BEHAVIOR IN SEATURTLES 43 Musick, J. A., and C. J. Limpus. 1997. Habitat utilization and migration in juvenile sea turtles. In P. L. Lutz and J. A. Musick (eds.), The Biology of Sea Turtles, pp. 137-164. CRC Press, Boca Raton, FL. Pritchard, P. C. H. 1976. Post-nesting movements of marine turtles (Cheloniidae and Dermochelyidae) tagged in the Guianas. Copeia 1976:749-754. Rhodin, A. G. J. 1985. Comparative chondro-osseous development and growth of marine turtles. Copeia 1985:752-771. Schmidt-Nielsen, K. 1997. Animal physiology: Adap­tation and Environment. 5th ed. Cambridge Univ. Press, Cambridge. Siegel, S., and N. J. Castellan. 1988. Nonparametric Statistics. McGraw-Hill Book Co., Inc. New York. Thompson, D., and M. A. Fedak. 2001. How long should a dive last? A simple model of foraging decisions by breath-hold divers in a patchy envi­ronment. Animal Behaviour 61:287-296. Van Dam, R. P., and C. E. Diez. 1997. Diving behavior of immature hawksbills (Eretmockelys imbricata) in a Caribbean reef habitat. Coral Reefs 16:133-138. Witham, R. 1977. Dermochelys coriacea in captivity. Marine Turtle Newsletter 3:6. ----------. 1980. The "lost year" question in young sea turtles. American Zoologist 20:525-530. Witherington, B. E. 2002. Ecology of neonate Logger­head Turtles inhabiting lines of downwelling near a Gulf Stream front. Marine Biology 140:843­853. Wyneken, J. 1997. Sea turtle locomotion: mechanisms, behavior, and energetics. In P. L. Lutz and J. A. Musick (eds.), The Biology of Sea Turtles, pp. 165­198. CRC Press, Boca Raton, FL. Wyneken, J., and M. Salmon. 1992. Frenzy and post­frenzy swimming activity in loggerhead, green, and leatherback hatchling sea turtles. Copeia 1992:478­484. Accepted: 5 October 2003. Journal of Herpetology, Vol. 38, No. 1, pp. 43-52, 2004 Copyright 2004 Society for the Study of Amphibians and Reptiles A New Species of Tantilla (Serpentes; Colubridae) of the Taeniata Group from Southern Belize Peter J. Stafford The Natural History Museum, Cromwell Road, London SW7 5BD, United Kingdom; E-mail: [e-mail] Abstract.-A new snake of the genus Tantilla is described from southern Belize. This species, a member of the taeniata group, is characterized by a dark gray-brown, almost black ground color; a narrow pale middorsal stripe confined to the vertebral scale row; a narrow pale lateral stripe on adjacent thirds of the third and fourth scale rows; a broad pale nape band that is complete medially; dark mottling on the lateral edges of the ventrals; and 153 ventrals + 64 subcaudals in the single known specimen, a female. It is most similar to Tantilla impensa of southern Chiapas, the central Guatemalan ranges and western Honduras but differs from this species in its darker overall color pattern, the presence of dark mottling on the lateral edges of the ventrals, and in having a lower number of ventrals. Resumen.-Se describe una nueva especie del genero Tantilla del sur de Belice. Esta especie pertenece al grapo taeniata y se caracteriza por lo siguiente: color oscuro gris-cafe, casi negro; una estrecha franja palida mediodorsal confinada en la fila vertebral de escamas; una estrecha franja palida lateral en los tercios adyacentes de las filas de escamas tercera y cuarta; una ancha banda palida en la nuca, completa medialmente; moteaduras oscuras en los bordes laterales de los ventrales y 153 ventrales + 64 subcaudales en la unica hembra conocida. Es similar a Tantilla impensa con una distribution en el sur de Chiapas y centro de Guatemala hasta el oeste de Honduras, de la que se diferencia por su color mas oscuro, presencia de moteaduras oscuras en los bordes laterales de los ventrales y por tener un menor numero de ventrales. In terms of its overall composition, the been the focus of several major inventory-based herpetofauna of Belize is relatively well known, studies, the first published more than 60 years Amphibians and reptiles of this country have ago (Schmidt, 1941), and a considerable wealth of44 PETER J. STAFFORD information has also accumulated on patterns of species distribution, local variation, and natural history (summarized in Lee, 1996,2000; Stafford and Meyer, 2000). Herpetological investigations in the Guatemalan department of El Peten, with which Belize shares its western border and has greatest faunal affinity, have been similarly extensive (see Campbell and Vannini, 1989; Lee, 1996; Campbell, 1998b). Given this history of attention it is probable that relatively few species of amphibians and reptiles remain to be discov­ered in the Belize/Peten area. Additional new forms, however, continue to emerge from time to time (e.g., Campbell et al., 1994; Mendelson, 1994; Campbell and Smith, 1997), and from a herpetological perspective, there are still many parts of the region that have yet to be explored. Small cryptozoic snakes are particularly nota­ble for their ability to escape detection, illustrated within the last few years by the discovery of several new species of Tantilla in the taeniata group (Campbell and Smith, 1997; Campbell, 1998a; Wilson et al., 1999). Of the 17 species currently recognized within this group (exclusive of the form described here), more than half have been described since 1971, and at least half remain known from fewer than a dozen speci­mens, providing some indication of the dif­ficulties in finding these snakes (Campbell and Smith, 1997). During recent fieldwork in the Vaca Plateau area of southwestern Belize, an­other snake of the Tantilla taeniata group was found that, although bearing clear resemblance to a species described from Guatemala, T. impensa (Campbell, 1998a), differs from that taxon in a number of prominent features. In attempting to determine the status of the specimen I compared it with representative material of impensa, as well as examples of other similar species within the taeniata group (see Appendix 1) and also con­sulted the detailed descriptions of these snakes provided by Campbell (1998a), Campbell and Smith (1997), Perez-Higareda et al. (1985), Savit­sky and Smith (1971), Wilson (1982,1983), Wilson and McCranie (1999), Wilson and Meyer (1971), and Wilson et al. (1999). The specimen cannot be allocated to any of the known species and is thus herein considered representative of a new form, which is described below. Materials and Methods Ventral count methodology follows that of Dowling (1951), and for ease of comparison with similar species, I have adopted the general sequence of character descriptions used by Campbell (1998a) and Campbell and Smith (1997). Abbreviations used are SVL (snout-vent length); TL (tail length); HL (head length, mea­sured from tip of snout to furthest edge of posterior-most supralabial); HW (head width, measured at angle of jaw); and ED (eye diameter, measured horizontally at is midpoint). All mea­surements except for SVL and TL were made to the nearest 0.1 mm using digital calipers held under a dissecting microscope. Sex was de­termined by observation of anatomical structure at the base of the tail through a small ventral incision. Dorsolateral and ventrolateral fields refer to the broad dark-colored areas between the narrow pale middorsal and lateral stripes, and the lateral stripes and ventrals, respectively. Affinities of the new form in relation to spe­cies with similar characteristics were assessed using nonmetric multidimensional scaling, based on data collected from specimens and selected literature records (Campbell, 1998a; Wilson and McCranie, 1999). Values for each character were standardized before analysis to Z-scores with a mean of 0 and standard deviation of 1, and the ordination of specimens along two NMDS dimensions was plotted. A two-dimensional NMDS solution was sought because the alterna­tive hypothesis suspected the existence of three similarity-based groupings. Tantilla hendersoni sp. nov. Figures 1-3 Holotype.-The Natural History Museum, London (BMNH) 2002.3 (field PJS 0242); a female from 0.5 km east of Las Cuevas on trail to Monkey Tail River, Cayo District (GR 1643'95"N,"8859'17" W), collected by P. Stafford, S. McMurry, and T. Rainwater on 13 September 2002. This locality (site of the Natural History Museum's field research station) lies in an area of semievergreen broadleaf forest at 580 m. Diagnosis.-A species of the T. taeniata group that may be distinguished from all other members of the genus by having (1) a dark gray-brown, almost black ground color; (2) a pale nape band that is complete medially; (3) a mid­dorsal stripe confined to the vertebral scale row and extending to the distal portion of the tail; (4) narrow pale lateral stripes occupying adjacent thirds of scale rows 3 and 4 that extend to the tip of the tail; (5) the lower third of the paraventral scale row mostly unpigmented; (6) dark mot­tling on the lateral edges of the ventrals; and (7) 153 ventrals and 64 subcaudals (excluding terminal spine) in the single known female. This species is compared with other regional mem­bers of the taeniata group in Table 1 and 2, and Figure 4. Description of Holotype.-A female with an SVL of 207 and TL of 65 mm (tail 23.9% of total length); body attenuate and gracile in form; HL 7.6 mm; HW 5.1 mm; head moderately distinct from neck; snout rounded to somewhat truncate in dorsal view; ED 0.9 mm, about 11.8% of headA NEW SPECIES OF TANTILLA 45 Fig. 1. Dorsal (top), ventral (center) and lateral (bottom) aspects of the head and forebody of BMNH 2002.3. Head length 7.6 mm from tip of snout to furthest edge of posterior-most supralabial. length and 35% of snout length; pupil subcircu­lar; rostral approximately 1.2 times broader than high; intemasals 2.3 times wider than long, laterally in contact with the anterior and poste­rior nasals; prefrontals large, slightly wider than long, laterally in contact with posterior nasal, second supralabials (right side only), and pre­ocular; median prefrontal suture 0.3 times as long as frontal; frontal approximately 1.5 times longer than wide; parietals 1.7 times longer than wide, median suture 0.6 of frontal length; nasals completely divided, nostril located mostly in posterior portion of anterior nasal; no loreal; 1/1 preoculars; 2/2 postoculars; temporals 1+1, separating supralabials 5-7 from parietal; supra­labials 7/7, first in contact with nasals, second in contact with postnasal and preocular, third in contact with preocular, third and fourth in contact with orbit, fifth and sixth in contact with anterior temporal, and seventh (the largest) in contact with anterior and posterior temporals; mental 1.4 times broader than long, in narrow contact with anterior pair of chinshields; anterior chinshields relatively large, about twice as long as wide; posterior chinshields about half of size of anterior chinshields, separated from first ventral by three gulars and two preventrals; infralabials 6/6, first four pairs in contact with Fig. 2. Midbody pattern characteristics of (from top to bottom) Tantilla hetuiersoni (holotype), Tantilla im­pensa (USNM 523955), and Tantilla taeniata (KU 289863).46 PETER J. STAFFORD Fig. 3. Hololype of Tantilla hcndcrsoni (BMNH 2002.3)/ a female 272 mm in total length; tail length 65 mm (Photograph© Frank Greenaway/The Natural History Museum). anterior chinshields, fourth pair largest; dorsal scales smooth, in 15 longitudinal rows through­out length of body; no apical pits evident; dorsal scales in 6 rows at level of tenth subcaudal; ventrals 153; anal divided; subcaudals 64 (ex­cluding terminal spine), paired. Ground color in preservative dark gray-brown, almost black; a contrasting pale middorsal stripe arising five scale lengths behind parietals, extend­ing to but less clearly delineated on tail (becoming obscure on distal third); middorsal stripe occupy­ing about median one-half of vertebral scale row; pale lateral stripes on upper third of scale row 3 and lower third of scale row 4, beginning about two scale lengths behind nape band and extend­ing to tip of tail; dark coloration on either side of the pale middorsal stripe (upper two-thirds of scale row 4 to lateral one-quarter of scale row 8) similar to that below pale lateral stripes (lower half of scale row 3, all of 2, and upper two thirds of row 1); lower third of scales on paraventral row pale; lateral edges of ventrals mottled with dark gray, this pigment extending beneath (and visible through) adjacent margins of para ventrals; venter of body and tail otherwise uniformly dull white; dorsal surface of snout pale brown, grading posteriorly to medium brown, with peripheries of major scutes and lateral/posterior edges of parietals darker; no pale spot on snout; complete palenapeband immediately behind parietalsand secondary temporals, nape band 2 dorsal scales long, slightly constricted medially, cream in color grading at level of venter to whitish; on side of head, an area of light pigment covering first and second supralabials, adjacent portions of third supralabial and posterior nasal, this marking continuous along lip margin beneath eye with similar postorbital "spot" involving fifth supra­labial, adjacent portions of fourth and sixth supralabials, lower postocular, and anterior tem­poral; circumorbital area blackish brown, this color extending ventrally as inverted fork-shaped mark on adjacent portions of supralabials 3 and 4 and posteriorly through upper postocular to merge with similar dark pigment on parietal; light postorbital marking and nape band separated laterally by dark gray-brown, extending to lip margin; infralabials immaculate, except for small grayish blotches along outer edge of mental and first two pairs of infralabials; venter of body and tail cream with dark gray-brown mottling on lateral edges of ventrals, this pigment appearing first on scale 13 and increasing in measure toward the tail. In life, the pale middorsal stripe was orange tan, the pale lateral stripes were yellowish white,A NEW SPECIES OF TANTILLA 47 Table 1. Characters used for multidimensional scaling analysisof specimens in the taeniata group of Tantilla, based on data collected from specimens and selected literature records (Campbell, 1998a; Wilson and McCranie, 1999). 1. Ventral scales 2. Subcaudal scales 3. Snout color (0 = not obviously paler than head; 1 = paler than head) 4. Condition of pale nape band (0 = complete; 1 = interrupted medially) 5. Condition of pale vertebral stripe (0 = confined to median area of vertebral scale row; 1 = inclusive of vertebral scale row and adjacent portions of paravertebral row) 6. Pale vertebral stripe bordered laterally with dark pigment (0 = yes; 1 = no) 7. Pale lateral stripe extending to distal portion of tail (0 = yes; 1 = no) 8. Pale lateral stripe bordered above by dark pigment (0 = yes; 1 = no) 9. Color of dark dorsolateral vs ventrolateral field (0 = distinctly paler; 1 = similar) 10. Dark mottling on lateral edges of ventrals (0 = present; 1 = absent) 11.Color of venter in life (0 = white; 1 = orange/ salmon) and the pale nape band was bright yellow, all of these markings contrasting sharply with the blackish brown color of the dorsolateral and ventrolateral areas. The dorsal surface of the head was pale brown anteriorly, grading on the posterior cranial scutes to medium brown, the peripheries of these scales notably darker; the light-colored areas on the side of the head were yellow; the mental, infralabials, and margins of adjacent chinshields were also yellow. Except for the dark mottling on the lateral edges of the ventrals, the venter was pearl white and semi- translucent. The tongue was dark brown at its tip, grading posteriorly to pale reddish brown. The iris of the eye was uniformly black. There are 14 stout, distally compressed teeth on the left maxillary. The anterior 12 are rela­tively small, increase in size posteriorly, and are separated from the last two (these somewhat enlarged) by a small diastema. Morphometric Analysis.-An ordination plot of character traits in 12 specimens of the T. taeniata group based on multidimensional scaling reveals the existence of three distinct clusters (Table 1, Fig. 4). These correspond to T. hendersoni and the two species it most closely resembles, T. impensa and T. taeniata. Separation along the first di­mensional axis relates mostly to differences between the specimens in snout color, width of the pale vertebral stripe, pattern characteristics of the dorsolateral and ventrolateral fields, presence/absence of dark mottling on the lateral edges of the ventrals, and coloration of the venter. The second dimension primarily sepa­rates hendersoni from the specimens of impensa. Separation along this axis is related to differ­ences in the number of ventral scales, coloration of the dorsolateral field, and presence/absence of dark mottling on the lateral edges of the ventrals. Etymology.-The specific name is a patronym in honor of Robert W. Henderson, Curator of Herpetology at the Milwaukee Public Museum, in recognition of his many published studies on the herpetofauna of Belize. Distribution.-Tantilla hendersoni is known only from the type locality. This lies in an area of distinctive karst topography noted for its many sinkholes and extensive underground cave sys­tems. Las Cuevas itself is named for its proximity to a large subterranean cavern. The limestone forests in the area are continuous with similar formations in the southern Vaca Plateau, Maya Mountains, and adjacent Peten region of Guate­mala, over parts of which the range of T. hender­soni is likely to extend. Ecological Notes.-Within the life zone classifi­cation system of Holdridge (1967), Las Cuevas lies in a transitional zone between Subtropical Wet and Subtropical Moist Forest. Vegetation in the area is characterized by seasonal broadleaf forest (Penn et al, 2003), with an approximate canopy height of 20-30 m and a relatively open understory predominated by vines and dwarf palms (Chamaedorea spp.). Common tree species include Brosimum alicastrum, Manilkara chicle, Dialium guianense, Cameraria latifolia, Sabal maur- itiiformis, Calophyllum brasiliense var. rekoi, and Coccololoba belizensis. Average rainfall for the type locality based on incomplete records available for the period 1995-1999 is estimated at 2650 mm per year (N. Garwood, unpubl. data), with most (> 70%) falling between June and December. No other member of the T. taeniata group is known from the collection site of T. hendersoni or elsewhere in southern Belize. To the northeast in the provincial districts of Corozal and Orange Walk, there are several records of T. cuniculator (Stafford and Meyer, 2000), and this species also occurs at localities in eastern Peten (Lee, 1996; Campbell, 1998b), but it appears to be an essentially lowland form (< 100 m) restricted to less mesic habitats. Given that hendersoni has remained unknown in the area for so long, despite the use of methods that have resulted in the detection of various other small leaf litterTable 2. Comparative features and distribution of snakes of the Tantilla taeniata group from Belize and eastern Guatemala. Includes data from Campbell (1998a), Wilson and McCranie (1999), and other selected literature records, "'based on an individual from Depto. Yoro, Honduras (UMMZ 58417); although higher than might normally be expected, Wilson and McCranie (1999) found no reason to consider this specimen as other than a representative of taeniata. 48 PETER J. STAFFORD o U & o •S ° 4, 3.2 w U -w ° s 2 •a g g f Ti o~ pg £h ps Cl, g £ o M-. f3 o is x C/j Qj Qj ^ <5 o > .5 o Qj £ CL, gS o U 6£ o ^.5 J » 2 o c o O > QJ ^ -a ,S is *, is J^T-5 . v. U C Vh « S a o a, §9-52 rS > y J" ~ )h T3 in U ft J=1 ^ S 5 O .5 TD > Cl, 'O >< S=1 pS £ O T5 . O £ M §" & Cl, qj £ C O "H u o u o u s QJ o tj o e C/j "pS CL, £ CO C o 'ft ^ o c D o c ^ o ^ Cl, *43 Qj C/j PS g s Ph u T5 c QJ e ' 8 w C5 f J J c -§ .5 « -£ "d ^ q3 C T5 ^ QJ O >< Qj PS tc jEl .3 Qj T3 u S QJ T5 C .3 QJ QJ T5 S=1 rS C ' o o T5 T5 c/j ri 13 JS Jh . __ CQ o| O O j5 B ® & SC'S ^ 8 -S 2 " o ^ ? s o Q- *H !> £ lH X> - o5 2 B.m s o H-c <8 o is X c/j <y QJ ^ be £ 7 > QJ ^ _ C QJ Qj tc S - *Ej c/j . £ g C/j1 *5 tc 'n ° "d 5P C »■ Qj U s J5 ns u e « o O fe X C/j Qj QJ f-1 be 7 > QJ ^ -S | ^ ' u 6 w 6C .3 QJ oS ^ a S 52 < c £ X -3 qj ti •K o Cl, C •£ h-h g T3 g cp ° o SN-0 ■" S QJ >H mB m -° 1) u ^ J=1 £S ^ ^ ps 5 ^3 O C 5 s s -SFo ^ T3 tc «+N C o ^ c/j q X -rt c I ■2^ ;d'- PS ^ cS ^ ‘ ! T3 C C O C c .2 © t; o o Xi > o X) 2 Qj <y -S y 'S *3 ^ PS PS f- Mi T3 cn QJ w PS ^ CL, j-j O Qj 413 o o s "o ' & 8 in PS z CL,*f ‘_d ro c/j C/j Cl, *S O ii >nlj w T3 X O O K Cl,x yf QJ Qj Qj X .&•§ & ^3X0 PS <y PS Ph >n PS D PS Q T5 ' QJ • e D c S SP 'a, c D tc ^ £ h ,£P ps o ^2 cu g 9-x; be C .2 w .2 3 tn pg Jh Cl,^> u JS '' ^ u 13 ^ O ^ c/j -3 Dx SP 'a, e D SP 'a, c D ti o >n in 5 SP ‘cl, e Qj T3 . PS QJ SPt3 ^ M ^ rr- O Jh S-h ■• & £ £ £ ps T5 ^ T5 u C o PS o c S QJ CL, p. f-t PS C Qj O kJ ? o c Cl,-^ o ■ B £ PS c 'Ts Ph pS c/5 s T5 SP QJ b3 -5 s-h pS O S QJ Q i 6C C fS .2 T5 T5 6£ T5 ^ f3 C s I J2 'o U r4 ( ( 04 4 vO vO dsci ro C/j C/j T3 . . PS S J3*Table 2. Continued. A NEW SPECIES OF TANTILLA 49 CM CO CM in cm cm <b O O N N N CM I I, l, in CO CM CM r-1 vO vO CM "sf m CM CM in CM In In d N III. co in r-; oj vO vO CM U CM CO CO cii 7 *? ^ o CO CO on CM in^r-CM .2 j2 OS 2 5 I s 2' 2$ QJ tXi 3 a M z M ca T5 C oj C W .3 ° in tft i-i 05 " OS Tl OS OS ^ « c g 3 o S .3 il -a Is ,^h Sh J> •SP 3 o rt ^ a e J? ■5 in T5 "73 S .y o Cl, CM O J-H CZD n m \ 2 - S <± 1 o o 05 ik^z 3 J5 ° § 2 t! 6 .££.3 U5 r* (IS (C 85 E S 9- 2 i! -a .« -g « s S r'S (S u o O E M (Xi %Jz- *05 13 hr 05 & c ° S *S Cl, & s, ■§- 3 Trt J2 £••- ^ m H £ S O > S 2 13 m c Jh ^ 13 13 jS a & - *3 ^ T5 Sh M OS P-i 0j &f oj S ,H s -5 °- -S Ji « 3 <fi § S H -t^, g S3 It:! |?& ■"Sis g 5 « u JJS ^s. Qj i> U5 "d g $ s^ .y 0J On >-h O 4J- o (Xi H 05 'O • »H ,i_i On 3 O ~ .a; O oj p u ^ 'Oh^ | | ca P P <2 ^-i in T5 -a P3 O Cl, CL> O O C be o T. impensa ° G * T. hendersoni (§> T. taeniata !? 8 ° t i I i i r -1.5 -1.0 -.5 0 .5 1.0 1.5 2.0 Dimension 1 Fig. 4. Ordination of specimens of Tantilla hender­soni (N = 1), Tantilla impensa (7), and Tantilla taeniata (4) based on results of multidimensional scaling analysis (Euclidean distance model). For details of characters used, see Table 1. snakes (intensive microhabitat surveys, drift fence and pitfall trap systems), it is likely to be a rare or at least uncommon species. The specimen was found crawling at the side of a muddy trail through dense forest at 2015 h, within a few meters of a small mussurana (Clelia clelia). A fecal deposit produced shortly after capture contained the fragmen­tary remains of a centipede. Discussion Comparisons.-Wilson and Meyer (1971) recognized two sections within the taeniata group of Tantilla based on differences between the species in dorsal color pattern, the reticulata section, consisting of Tantilla flavili- neata, Tantilla oaxacae, and Tantilla reticulata and the taeniata section (subsequently rede­fined by Wilson, 1983), to which Tantilla briggsi, Tantilla cuniculator, T. impensa, Tantilla jani, Tantilla johnsoni, Tantilla slavensi, Tantilla striata, T. taeniata, Tantilla tayrae, Tantilla tecta, Tantilla tritaeniata, Tantilla trilineata, Tantilla triseriata, and Tantilla vulcani can be allocated. In having a pale lateral stripe on scale rows 3 and 4, T. hendersoni is associated nominally with the latter section; the pale lateral stripe in T. flavilineata and T. oaxacae is wider, occupy­ing scale row 4 and adjacent portions of rows 3 and 5, whereas in T. reticulata scale row 4 and adjacent portions of rows 3 and 5 are brown. Tantilla flavilineata and T. oaxacae further differ from T. hendersoni in having fewer subcaudals in both sexes (43-52 and 46-52, respectively, vs. 64). Tantilla hendersoni differs from all other species in the T. taeniata group except T. impensa, T. jani, T. slavensi, and T. tecta in having a narrow pale middorsal stripe con-50 PETER J. STAFFORD fined to the central area of the vertebral scale row. In T. briggsi, T. cuniculator, T. tayrae, and T. vulcani, the pale middorsal stripe is absent, ill-defined posteriorly, or reduced to a series of spots, whereas in the remaining species it is expanded laterally to include all of the vertebral scale row and adjacent portions of the paravertebral rows. The pale lateral stripe in T. impensa, T. jani, T. slavensi, and T. tecta is distinctly bordered above (and in T. jani, T. slavensi, and T. tecta also below) by dark flecking or a narrow, continuous dark line, and in these species the color of the dorsum either side of the middorsal stripe is also somewhat paler. Tantilla jani may be further distinguished from T. hendersoni in having less distinct pale lateral stripes that usually terminate on the posterior section of the body, a nape band that is reduced to a pair of pale spots, a smaller light postorbital marking confined mostly to the fifth supralabial, the first pair of infralabials usually in contact, and fewer segmental counts (females with 143-147 ventrals and 44-47 sub- caudals vs. 153 + 64 in the female holotype of T. hendersoni). Tantilla slavensi further differs from T. hendersoni in having a pale nape band that is interrupted medially (vs. complete) and no more than one dorsal scale in length (vs. two in T. hendersoni), a pale lateral stripe that is in­distinct on the tail (vs. evident even at tip of tail), an orange venter in life (vs. pearl white), and 52­56 subcaudals (vs. 64 in T. hendersoni). In T. tecta, the snout is marked with a pale spot involving the upper portion of the rostral, internasals, and anterior two-thirds of the prefrontals, the mid­dorsal pale stripe is ill-defined posteriorly, and there are fewer ventrals and subcaudals (148 + 54 in the single known female). Ventral and subcaudal scale counts in T. hendersoni fall directly within the range of T. taeniata, to which the new form also bears resemblance in aspects of proportion and general habitus. As indicated by the ordination pattern in Figure 4, however, it is perhaps most closely related to T. impensa. This species has a compara­ble number of subcaudals (65-72 in females vs. 64 in the holotype of T. hendersoni), a similarly narrow middorsal stripe that is confined to the central area of the vertebral scale row and extends onto the distal portion of the tail, and a dark ventrolateral field. In T. impensa however, the dark ground color on either side of the middorsal stripe is typically paler than that below the lateral stripe (Fig. 2) and bordered above and below by dark gray-brown, the lateral edges of the ventrals are unpigmented, and females have a higher number of ventrals (164­172 vs. 153 in T. hendersoni). Tantilla impensa is potentially also a larger species, with females attaining total lengths in excess of 720 mm (Campbell, 1998a). As presently defined, the range of T. impensa extends from eastern Chiapas, Mexico, across the northern slopes and foothills of eastern Guate­mala (Montanas del Mico, Sierra del Merendon, Sierra del Esplritu Santo to northwest portion of Sierra de Los Cuchumatanes) to western Hon­duras (Wilson and McCranie, 1999), thus placing it within considerable geographic proximity to the new form described here. The two species also appear similar in their ecological distribu­tion, T. impensa having been recorded from Tropical and Subtropical Wet forest formations at elevations from near sea level to 1600 m. The evident distinctions between T. impensa and T. hendersoni in color pattern and ventral scale numbers, however, suggest that they are not simply representatives of different populations of the same species, a conclusion further supported by the fact that geographic variation in T. impensa appears to be low; throughout its range, the pale nape band in T. impensa may be either complete or interrupted dorsally (Campbell, 1998a; Wilson and McCranie, 1999), and in SMF 79114 from Depto. de Copan, Honduras (a small juvenile with lower body and tail missing), the dorsolat­eral coloration is exceptionally dark, but in the nature of striping and all other diagnostic elements of color pattern this species appears to be relatively stable. Four additional species of Tantilla occur in the area of Belize with which T. hendersoni is broadly sympatric and may be confused; Tantilla cunicu­lator, Tantilla moesta, Tantilla schistosa, and Tantilla tecta. Tantilla cuniculator has a brown dorsum with usually no evidence of a middorsal stripe, a pale nape band that is orange-brown in color and involves the posterior tips of the parietals, a pale yellow-orange spot on the snout, a reddish orange venter, and fewer subcaudals (48-53). Tantilla moesta has a longer pale nape band that extends posteriorly from the parietals for a length of at least three dorsal scales and usually includes the posterior-most three supralabials. In this species, the dorsum and venter are also uni­formly dark brown or black, and there are no longitudinal stripes. Tantilla schistosa has a uni­formly olive-colored or reddish brown dorsum, a pale nape band that involves the posterior portions of the parietals, 20-40 subcaudals, and a relative tail length of 12.8-20.1% (vs. 23.9% in the holotype of T. hendersoni). Tantilla tecta is known from a single locality in northeastern Peten (Campbell and Smith, 1997) and differs in having a pale spot on the snout, enclosing the upper portion of the rostral, internasals and anterior two-thirds of the prefrontals, a pale middorsal stripe that is indistinct on the lower body, and 148 ventrals and 54 subcaudals in the single known female. Tantillita canula and Tantil- lita lintoni also occur in Belize, of which T. lintoniA NEW SPECIES OF TANTILLA 51 has been further reported from the same specific locality (Stafford and Meyer, 2000). These species are easily distinguishable from T. hendersoni, however, in lacking a pale nape band and lateral stripes, and in having the middorsal stripe (if present) usually reduced to a series of spots posteriorly. Several species of Tantilla to which T. hendersoni is ostensibly related are known from only a few specimens or, as in the similar case of T. tecta, only the holotype. Consequently, there is no possibil­ity of assessing variation in the traits deemed to be distinctive of these forms, a number of which resemble one another closely and have been described from a fairly limited area. But the alternative-to wait until comparative material becomes available-is not practical when dealing with such infrequently collected (and possibly rare) species. Clearly, a pressing need exists for detailed systematic studies on the T. taeniata group of Tantilla, and molecular techniques may be of particular benefit in elucidating relation­ships among these snakes. Acknowledgments.-For the loan of compara­tive material I am indebted to A. Resetar and J. Ladonski (Field Museum of Natural History), G. Kohler (Forschungsinstitut und Naturmuse- um Senckenberg), R. Henderson and G. Casper (Milwaukee Public Museum), J. Simmons (Museum of Natural History, University of Kansas), G. Zug and R. Wilson (Smithsonian Institution, United States National Museum), J. Campbell and P. Ustach (University of Texas at Arlington), and C. McCarthy (The Natural History Museum, London). In Belize, C. Minty, N. Bol, and E. Saquil of the Las Cuevas Research Station assisted with practical aspects of field­work, and for their companionship during many hours of trail walking, I extend thanks to S. McMurry and T. Rainwater (Texas Tech Univer­sity, Lubbock). Fieldwork was undertaken with funding assistance from the Natural History Museum (London) and the British Ecological Society, and for permission to work in the Chiquibul Forest Reserve grateful thanks are also due to the Ministry of Natural Resources (Belize). The Spanish summary was kindly prepared by M. Pena, and for advice on procedures of character analysis I am indebted to N. MacLeod and D. Rose (Natural History Museum, London). Helpful review comments on the original draft of this manuscript were received from L. D. Wilson (Miami-Dade Com­munity College) and an anonymous reviewer. Literature Cited Campbell, J. A. 1998a. Comments on the identities of certain Tantilla (Squamata: Colubridae) from Gua­temala, with the descriptions of two new species. Scientific Papers, Natural History Museum, Univ. of Kansas 7:1-14. ----------. 1998b. Amphibians and Reptiles of Northern Guatemala, the Yucatan, and Belize. Oklahoma Univ. Press, Norman. Campbell, J. A., akd E. N. Smith. 1997. A new species of Tantilla (Serpentes: Colubridae) from northeastern Guatemala. Proceedings of the Biological Society of Washington 110:332-337. Campbell, J. A. akd J. P. Vakkiki. 1989. Distribution of amphibians and reptiles in Guatemala and Belize. Proceedings of the Western Foundation of Verte­brate Zoology 4:1-21. Campbell, J. A., J. M. Savage, akd J. R. Meyer. 1994. A new species of Eleutherodactylus (Anura: Leptodac- tylidae) of the rugulosus group from Guatemala and Belize. Herpetologica 50:412-419. Dowlikg, H. G. 1951. A proposed system for counting ventrals in snakes. British Journal of Herpetology 1:97-99. Holdridge, L. R. 1967. Life Zone Ecology. Tropical Science Center, San Jose, Costa Rica. Lee, J. C. 1996. The Amphibians and Reptiles of the Yucatan Peninsula. Comstock Publishing Associ­ates, Ithaca, NY. ----------. 2000. A Field Guide to the Amphibians and Reptiles of the Maya World: The Lowlands of Mexico, Northern Guatemala, and Belize. Com­stock Publishing Associates, Ithaca, NY. Mekdelsok III, J. R. 1994. A new species of toad (Anura: Bufonidae) from the lowlands of eastern Guatemala. Occasional Papers, Natural History Museum, Univ. of Kansas 166:1-21. Pekk, M. G., D. A. Suttok, akd A. Mokro. 2003. Vegetation of the Greater Maya Mountains, Belize. Systematics and Biodiversity. Perez-Higareda, G., H. M. Smith, akd R. B. Smith. 1985. A new species of Tantilla from Veracruz, Mexico. Journal of Herpetology 19:290-292. Savitsky, A. H., akd H. M. Smith. 1971. A new snake from Mexico of the taeniata group of Tantilla. Journal of Herpetology 5:167-171. Schmidt, K. P. 1941. The amphibians and reptiles of British Honduras. Field Museum of Natural History Publications, Zoology Series 22:475-510. Stafford, P. J., akd J. R. Meyer. 2000. A Guide to the Reptiles of Belize. Academic Press, San Diego, CA. Wilsok, L. D. 1982. A review of the colubrid snakes of the genus Tantilla of Central America. Milwaukee Public Museum Contributions in Biology and Geology 52:1-77. ----------. 1983. A new species of Tantilla of the taeniata group from Chiapas, Mexico. Journal of Herpetol­ogy 17:54-59. Wilsok, L. D., akd J. R. McCrakie. 1999. The systematic status of Honduran populations of the Tantilla taeniata group (Serpentes: Colubridae), with notes on other populations. Amphibia-Reptilia 20:326-329. Wilsok, L. D., akd J. R. Meyer. 1971. A revision of the taeniata group of the colubrid snake genus Tantilla. Herpetologica 27:11-40. Wilsok, L. D., R. K. Vaughak, akd J. R. Dixok. 1999. Another new species of Tantilla of the taeniata group from Chiapas, Mexico. Journal of Herpetology 33:1-3.52 PETER J. STAFFORD Appendix 1 Specimens Examined Tantilla cuniculator. Belize: Tower Hill, Orange Walk District (MPM 7608). Mexico: 9.7 km SE Coba, Quintana Roo, 0 m (KU 171745). Tantilla impensa: Honduras: Quebrada Grande, Depto. de Copan, 1600 m (FMNH 236413; USNM 523955); Quebrada Grande, Depto. de Copan, 1300 m (SMF 79114); Lago de Yojoa, Depto. de Accepted: 31 October 2003. Cortes (USNM 523956). Guatemala: Livingston, Punta Cocoli, Izabal, 2 m (UTA 39550, paratype); Los Amates, Aldea Vista Hermosa, Izabal, 650 m (KU 191103); Los Amates, Sierra del Espiritu Santo, approximately 1 km NE Aldea San Antonio, Izabal, approximately 660 m (UTA 28532, paratype). Tantilla jani: Guatemala: no specific locality (BMNH 1946.1.8.68, lectotype). Tantilla taeniata: El Salvador: Usulutan, Cerro del Tigre, 1100 m (KU 289863). Guatemala: no specific locality (FMNH 40890); Parque Nacional Laguna El Pino, Santa Rosa (UTA 22848). journal of Herpetology, Vol. 38, No. 1, pp. 52-60, 2004 Copyright 2004 Society for the Study of Amphibians and Reptiles A New Species of Echinosaura (Squamata: Gymnophthalmidae) from Ecuador Gunther Kohler,1'2 Wolfgang BOhme,3 and Andreas Schmitz4 1 Forschungsinstitut und Naturmuseum Senckenberg, Senckenberganlage 25, D-60325 Frankfurt a. M., Germany; E-mail: [e-mail] 3Zoologisches Forschungsinstitut und Museum Alexander Koenig, Ademuerallee 160, D-53113 Bonn, Germany 4Museum d'Histoire Naturelle, Route de Malagnou, CH-1211 Geneve 6, Switzerland Abstract.-A new species of Echinosaura, Echinosaura brachycephala, is described from two localities on the Pacific versant of the Ecuadorian Andes. The new species differs from all other species of the genus by its conspicuously short head with a high domed snout in lateral profile and various scalation characteristics including the number of ventral scales per caudal segment, the arrangement of dorsal body scales, and a reduced or absent postmental scale. A key to the species of Echinosaura and Teuchocercus is provided. Lizards of the genus Echinosaura are small terrestrial "microteiids" that are distributed from western Panama across northwestern Colombia to west-central Ecuador. Although usually found in the vicinity of streams, these lizards live somewhat more terrestrially than the semiaquatic Neusticurus (Uzzell, 1965a,b). Although they appear to be predominantly nocturnal (Uzzell 1965a), active Echinosaura can be observed during the day as well (Dunn, 1944; Breder, 1946; Uzzell, 1965a). In his revision of Echinosaura, Uzzell (1965a) recognized a single species (Echinosaura horrida) divided into three subspecies. A few years later, a new genus and species, Teuchocercus keyi, presumably related to Echinosaura, was described by Fritts and Smith (1969). Recently, Fritts et al. (2002) have described a new species of 2 Corresponding Author. Echinosaura and have elevated all three sub­species of E. horrida to species. Thus, as currently understood, the genus Echinosaura contains four species: E. horrida Boulenger, Echinosaura orcesi Fritts et al., Echinosaura palmeri Boulenger, Echi­nosaura panamensis Barbour. We examined a rather large series of an apparently undescribed, conspicuously short­headed species of Echinosaura from two localities on the Pacific versant of the Ecuadorian Andes. The differences in head shape and some scalation characteristics are so conspicuous that this taxon cannot be confused with any of the known species of Echinosaura. Materials and Methods A list of the specimens examined is provided in Appendix 1. Additional data for E. orcesi were taken also from Fritts et al. (2002). Discriminant function analysis (performed with Statistica forTable 1. Selected measurements, proportions and scale characters in Echinosaura brachycephala (10 males, 15 females), Echinosaura orcesi (1 male, 1 female; data also from Fritts et al. 2002), Echinosaura horrida (14 males, 13 females), Echinosaura palmeri (6 males, 5 females), Echinosaura panamensis (5 males, 5 females), and Teuchocercus keyi (6 A NEW SPECIES OF ECHINOSAURA 53 50^2 c, n II W . 1 <u s ts a 5s 5 flj tt O Cu QJ X) . • t/j t/j Ih e « 0 «4H xl o ns T £ £ ii 1 3 cfi 12 <y »h QJ ? ° 2 fe S « •S s s ra .3 2 a "S > ts QJ Qj T5 > T5 2 "d J7 C ns nS ctj t/j o qj "d n £ o Qj *§ I * X 2 TJ & £ O bfc 50^ % M § > g i. <in ° c tv CTj 2 >• 3 ccn c p p vD O 00 00 + 1 m +1 ! ' in ■ ' rL 1 ON ' ' r# ' o o ^ , I ,1 tv +1 tv +1 o o QkQk Jqiq tv p tv p T5 OJ T5 '> tj >-, s=i ns .3 T5 a 5b e o ro O O O r1 r1 o o o O 00 . , tv , . 00 +1 00 +1 o +1 o +1 o G> ^CJN^rO 9®9oo 7^7^ A 3 >1 3 w w 0^0^ rH r-> O O ; Os H vD O O o K tv vD 00 O o as + +1 vD ' tv 00 o CM O o o as +1 O O tv ; CM 00 ^ tv tv ph ns £ ns 3*8 £ > cn +i m o o +1 +1 t t 7 : mm oo ■ T5 QJ T5 IS tj >-, s=i ns .3 T5 ‘5b e o ■s: T5 ' ;g ! TJ >■» ‘ in m o o P P o o o o ^ ON n ,1 . . ON +1 00 +1 vO +1 O +1 G> O ^ as o oo ^ ; vO 00 ^ 00 T5 QJ T5 IS tj .3 T3 a t' o • 00 ON Q tv Q tv tv O tv O o w o w o +1 "5b Tc ' .3 -3 1 QJ QJ O p "5b Ik r-5 rH .3 CTj .3 CTj ns u ns u CM O o o o > +1 l-^1 ^1 ~ ■ m 7 $ 7 S i h vD O vD O < w o w o w +1 +1 . . CM N m tM Qj Qj _• O _• rH ,-25 ,-25 o C o bo bcToTo G G p w p w Vj Vj o o C 42 o £ J-H O pH o o +1 +1 rn CM m m 11 cm cm m o O in tv m vD o o o o o O o o + 1 +1 +1 + 1 +1 + 1 CM CM CM o o o K o s S in in °P j U|> U|> in in tv o m m tv p O ON vD vO in CM O O o o rH rH + 1 + 1 + 1 +1 + 1 + 1 m CM in vD 00 O s s K O in in in in 00 00 vD 4 4 4 <k Jo K ^ in o + 1 O o o o o o + 1 +1 +1 o o o vQ in tv +1 +1 +1 m m 00 p o o t m t m ON ! 00 w o o o o m mm i vo ! ! 4 ! CM CM rH in o vD 00 CM vD o o O O o O o o + 1 + 1 +1 + 1 +1 +1 CM m 00 tv p p 00 ci m s m m T 7 ir> m m tv CM <y ns <y ns aj ns 2i o Cu X ns T5 S ns u "§ CD54 G. KOHLER ET AL. □ o DOa^ o • °‘a*8W> ° O oo 0 °0? • o °o A ■ A O A A *<>A A A O 0 O O O 0 O o Fig. 1. Discriminant function analysis based on three characters; empty circles: Echinosaura brachycc- phala; empty squares: Echinosaura horrida; solid squares: Echinosaura orccsi; diamonds: Echinosaura palmeri; tri­angles: Echinosaura panamcnsis, and solid circles: Tcuchocercus keyi. See text for details. Windows vers. 6) was used to evaluate the phenetic distinctness of a priori groups (see Appendix 1 for specimens used in this analysis). Nomenclature of dorsal head scales follows Fritts et al. (2002). All measurements were made using precision calipers and were rounded to the nearest 0.1 mm. Head length was measured from the tip of the snout to the anterior margin of the ear opening. Snout length was measured from the tip of the snout to the anterior border of the orbit. Head width was determined as the maximum width of the head. The number of ventral scales per caudal segment (in a longitudi­nal row) was determined at the seventh caudal segment. Abbreviations for museum collections follow those of Leviton et al. (1985). Results A comparison of selected morphometric and scalation characters in the short-headed Echino­saura, E, horrida, E, orcesi, E, palmeri, and E. panamensis, is provided in Table 1. Nonoverlap­ping differences between the short-headed Echi­nosaura and the remaining species were observed in the number of ventral scales per caudal segment and the relative size of the postmental scale. Only slightly overlapping ranges were found in the relative snout length (ratio snout length/snout-vent length). Data from 25 specimens of the short-headed Echinosaura, 27 specimens of E. horrida, 2 speci­mens of E. orcesi, 11 specimens of E. palmeri, 10 specimens of E. panamensis, and 3 specimens of T. keyi were included in a discriminant function analysis (DFA). Males and females were ana­lyzed together because no statistically significant differences were detected between sexes in the a. Fig. 2. Head of holotype of Echinosaura brachycc- phala (1VIHNG 2359.77). (A) Lateral view; (B) dorsal view; (C) ventral view. Scale bars equal 5.0 mm. included characters. Figure 1 shows the results of a DFA based on the following characters: (1) POSTMENTAL: level of first infralabial to which the postmental reaches (if the postmental did not reach to first infralabial or was absent, it was recorded as zero); (2) SNOUT: ratio snout length/snout-vent length (SVL); and (3) PARAVERTEBRALS: number of scales between paravertebral longitudinal rows of tubercles on dorsum. The first discriminant function (eigen­value 21.357) was DS = 0.512 [SNOUT] + 0.947 [POSTMENTAL] - 0.607 [PARAVERTEBRALS], The second discriminant function (eigenvalue 7.055) was DS = -0.447 [SNOUT] - 0.087 [POSTMENTAL] - 0.909 [PARAVERTEBRALS], The first and second discriminant functions correctly classified 100% of the Ecuadorian a priori groups but failed to completely separate the specimens of E. orcesi, E. palmeri, and E. panamensis. We describe the short-headed Echi­nosaura as follows. Echinosaura brachycephala sp. nov. Holotype.-MHNG 2359.77 (Fig. 2), an adult male, from Las Pampas (= San Francisco de lasA NEW SPECIES OF ECHINOSAURA 55 Pampas, 0°25'60"S, 78°58'0''W/1275 m elevation), Provincia Cotopaxi, Ecuador, collected May 1985 by G. Onore. " Paratypes.-MHNG 2359.26, 2359.34-37, 2359.43-44, 2359.51, 2359.53, 2359.56, 2359.62­63, 2359.65-66, 2359.68-72, 2359.74, 2359.76-78, 2359.80,2359.83-87,2359.91,2359.93-94,2359.98, SMF 81603-04, ZFMK 46370-75, 76376, same collecting data as holotype; MHNG 2360.1, 2360.4A, 2360.11, 2360.14, 2360.28, 2360.30, 2360.34, SMF 81605, from Tandapi (= Manuel Cornejo Astorga, 0°25'0''S, 78°47'60''W, 1665 m elevation), Provincia Pichincha, Ecuador, collect­ed October 1984 by G. Onore; 2360.39, from Tandapi, Provincia Pichincha, Ecuador, collected December 1983 by G. Onore. Most paratypes are adults, except MHNG 2359.70-72, 2359.84-85, 2359.87, 2359.91, 2359.93-94, 2360.1, 2360.4A, 2360.28,2360.30, and 2360.34, which are juveniles and subadults (SVL less than 55 mm). Diagnosis.-A medium-sized species (SVL in largest specimen 78.0 mm) of the genus Echino­saura that differs from all other species of this genus by the number of ventral scales per caudal segment (four scales per caudal segment in E. brachycephala vs. three scales in the other species (except E. orcesi which has 5-6 scales per caudal segment) and the relative size of the postmental scale (postmental scale reduced or absent in E. brachycephala, not reaching beyond one-third of first infralabial vs. postmental not reduced, always reaching well beyond one-half of first infralabial in the other species). It differs also by its relatively short snout (ratio snout length/SVL 0.06-0.07 in E. brachycephala vs. 0.07-0.09 in the other species). It can be readily distinguished from E. horrida by its dorsal scalation (2-6 scales between longitudinal rows of tubercles in E. brachycephala vs. longitudinal rows of tubercles juxtaposed in E. horrida) and the absence of a conspicuous pale band across chin (present in E. horrida). Echinosaura brachycephala is further differentiated from E. palmeri and E. panamensis by having a single internasal (divided in E. palmeri and E. panamensis) and the number of infralabials (three in E. brachycephala vs. 4-5 in E. palmeri and E. panamensis). Echinosaura brachycephala can be further distinguished from Echinosaura orcesi by having a lower femoral pore count (males: 7-9 in E. brachycephala vs. 14-15 in E. orcesi; females: 1-2 in E. brachycephala vs. 5-6 in E. orcesi). Also, E. orcesi lacks continuous rows of tubercles in the paravertebral area of the dorsum (present in E. brachycephala). Echinosaura brachycephala differs from the only species in the presumably closely related genus Teuchocercus, T. keyi, by its different caudal scalation (tail with only small conical scales in E. brachycephala vs. tail bearing conspic­uous whorls of spines in T. keyi), the number of ventral scales per caudal segment (four scales per caudal segment in E. brachycephala vs. three scales in T. keyi) and by the absence of granular scales covering all of the tympanum (present in T. keyi although this is variable with adults having more extensive coverage than juveniles but always discernable; in some E. brachycephala there are granular scales in the periphery of the tympa­num). Also, most T. keyi have 2-4 scales separating the paravertebral rows (usually 4-6, rarely two, in E. brachycephala), usually paired internasals (sin­gle in E. brachycephala) and some individuals have prefrontals (absent in most E. brachycephala). Most individuals of T. keyi have pale markings on the base of the tail (absent in adult E. brachycephala but usually present in juveniles). Description of the Holotype.-Adult male, as indicated by swollen base of tail and number of femoral pores; SVL 68.5 (all measurements in millimeters); tail (complete) length 112.0; axilla to groin distance 29.8; head length 16.8; snout length 4.6; head width 11.4; shank length 10.8; tongue with imbricate scalelike papillae; head scales mostly smooth, some wrinkled or tubercu- late and convex, none with numerous longitudi­nal ridges; rostral scale wider (2.8) than long (1.2), higher than adjacent supralabials, in contact with internasal, nasal and first supralabials posteriorly; internasal single, wider (2.0) than long (1.5), posterior suture angular with point directed posteriorly, in contact with nasals laterally, fron- tonasals posteriorly; nostril pierced in a single large nasal; nasal posteriorly in contact with frenocular (subrectangular, length 1.1, height 1.3) and lorilabial (subtriangular, length 1.7, maximum height 1.5); a pair of frontonasals (length 3.5, width 1.6), in contact with each other medially (suture length 2.5), in contact with presupraocular scales, frontal posteriorly, and frenocular laterally; frontal single, subhexagonal, longer (2.5) than wide (2.3), anterior suture angular with point directed anteriorly, lateral sutures almost straight, posterior suture angular with point directed posteriorly, in contact with presupraocular scales laterally, first supraocular on left side, frontoparietals posteriorly; a pair of frontoparietals (length 1.9, width 1.0), subrectan­gular, in contact with each other medially (suture length 1.3), in contact with first and second supraoculars; supraoculars three, third supra­ocular much smaller than first and second, only the first supraocular in contact with ciliaries; superciliary series complete; transparent area (palpebral disc) in lower eyelid unpigmented, divided by vertical grooves into 3 parts; parietal, interparietal, occipital, and temporal areas with small to medium-sized, polygonal, convex to tuberculate scales; tympanum superficial, pig­mented; only slightly recessed in an external auditory meatus; supralabials four; infralabials three; mental wider (3.5) than long (1.5), in contact56 G. KOHLER ET AL. with first infralabials, postmental posteriorly; scales on ventral surface of head irregular; postmental wider (2.5) than long (1.4), subpen­tagonal, posterior suture angular, point directed posteriorly, in contact with first infralabial; genials not differentiated; no gular fold; dorsal and lateral neck scales heterogeneous, conical scales inter­spaced with smaller, granular scales, cones of upper sides of neck in more or less longitudinal rows; ventral neck scales heterogeneous, larger keeled to low conical scales interspaced with smaller, granular scales; dorsal scales heteroge­neous, large tuberculate scales variously dis­posed, interspaced with smaller, almost granular scales; a pair of paravertebral rows of tubercles, slightly undulating on posterior portion of body, separated by a minimum of two scales in thoracic region, separated by 4-6 scales on most of dorsum; most of ventral scales arranged in obtusely keeled longitudinal and transverse rows; complete transverse ventral scale rows 23 (be­tween levels of axilla and groin); longitudinal ventral scale rows at midbody 8; cloacal plate scales four, medial two much larger than lateral scales; four rows of scales between posteriormost ventral plates and cloacal plates; tail cyclotetrag- onal, slightly compressed; dorsal and lateral surface of tail with enlarged conical tubercles; these tubercles are scattered irregularly at the base of tail but are arranged regularly (but both longitudinally and transversally disjunct) begin­ning distal to base of tail where segments are discernable; largest tubercle at the end of each segment, preceded by two gradually decreasing scales in a longitudinal line, but ending 2-3 scale rows before the anterior margin of the preced­ing caudal segment; anteriormost scales in each caudal segment about the same size as adjacent scales; ventral caudal scales smooth, flat, with 4 subcaudals in one segment (in a longitudinal row). Limbs pentadactyl; digits clawed; forefoot without enlarged platelike scales along inner margin between thumb and wrist; dorsal and ventral brachial scales polygonal, of varying sizes, keeled to tuberculate; dorsal manus scales polygonal, smooth, subimbricate; palmar scales small, polygonal, smooth; dorsal scales on fingers single (except on proximal portion of fourth finger) smooth, quadrangular, covering dorsal half of digit, overhanging subdigital scales, four on I, seven on II, 10 on III, 12/11 on IV, six on V; subdigital scales smooth, in single series, 6/7 on I, 11 on II, 17/16 on III, 16 on IV, 10/12 on V; anterodorsal thigh scales heteroge­neous, enlarged keeled scales interspaced with smaller, granular scales; ventral thigh scales more or less homogeneous, smooth, subimbricate; femoral pores 8/9; six scales between medial- Fic. 3. Variation in dorsal head scales in the type series of Echinosaura bracln/cepliala. (A) MHNG 2359.63; (B) MHNG 2359.26; (C) MHNG 2359.98; (D) MHNG 2360.14; (E) MHNG 2359.37; (F) SMF 81603. Fronto- nasals with light shading and frontals with dark shading. most femoral pores; anterior and anteromedial pes scales polygonal, subimbricate, smooth, irregular in size with the largest scales distally; scales on dorsal surface of digits single (except on proximal portions of toes III-V), quadrangular, smooth, overhanging subdigital scales, 5/4 on I, 8/7 on II, 15 on III, 24/23 on IV, 14 on V; subdigital scales single, eight on 1,13 on II, 20/19 on III, 26 on IV, 19 on V. Coloration in Preservative (70% Ethanol).-Dor­sal surfaces of head, body and tail uniform; lateral body brown with six (right) and four (left) vague pale brown round blotches; lateral head dark brown with some vertical pale brown streaks reaching onto labials; lower surface of head pale brown with a few dark brown spots; belly irregularly checkered with dark and pale brown; ventral surfaces of limbs and tail paleA NEW SPECIES OF ECHINOSAURA 57 Fig. 4. Lateral view of head. (A) Echinosaura horrida (ZFMK 43768); (B) Echinosaura brachycephala (ZFMK 46371); (C) Teuchocercus keyi (MHNG 2284.59); (D) Echinosaura palmeri (BMNH 1923.10.12.14); (E) Echinosaura panamensis (ZFMK 49107); (F) Echinosaura orcesi (NMW F 3087). Drawings by Mathias Gunther. brown with dark brown mottling; palmar and plantar surfaces pale brown. Variation.-The paratypes agree well with the holotype in all characters. Most variation is observed in dorsal head scalation (Fig. 3). Although all specimens have a single intemasal and a pair of frontonasals, there is considerable variation in the frontal and prefrontal region. Twelve of 41 specimens (29.3%) have the frontal longitudinally divided. There is a single unpaired scale centrally between the frontonasals and the frontal in six of 41 specimens (14.6%). Two of these specimens have an additional scale between the frontal and the frontoparietals. In three of 41 specimens (7.3%), there are 1-3 irregular scales ("prefrontals") between the frontonasals and the frontal, and in two specimens, there are three relatively symmetri-58 G. KOHLER ET AL. A NEW SPECIES OF ECHINOSAURA 59 Fig. 6. Distribution of the Ecuadorian species of Echinosaura. Echinosaura brachycephala (triangles; type locality encircled), Echinosaura horrida (circles), Echino­saura orcesi (squares), and Teuchocercus keyi (inverted triangles). Open symbols represent literature records (Fritts et al., 2002). Pale shading indicates elevations above 1000 m; darker shading indicates elevations above 2000 m. cal scales ("prefrontals") between the fronto- nasals and the frontal. In one specimen (SMF 81603), the internasal is fused with the left nasal (Fig. 3F). The range for relative tail length (ratio tail length/SVL) is 1.62-1.80. Supralabials are 3-5 (mostly 4), infralabials are constantly three. Subdigital lamellae on the fourth finger are 16-23, on the fourth toes 23­32. See Table 1 for variation in selected measurements and proportions and scale char­acters in the type series. Variation in coloration is minimal. Most juveniles and subadults (SVL less than 55 mm) have a more or less conspicuous pair of pale brown blotches on the dorsal surface of the base of tail. These pale markings obviously disappear with age, and only two of the adults examined (MHNG 2359.76, SVL 62.0 mm; MHNG 2359.65, SVL 60.0 mm) have more or less distinct pale brown blotches at the base of tail. The smallest specimen examined is MHNG 2359.71 (SVL 34.0 mm). Etymology.-The species name is derived the Greek brachy, meaning "short" and the Greek cephal, meaning "head" in reference to the con­spicuously short head of this species. Discussion The relationships of E. brachycephala remain unclear. Its head shape (especially in lateral view) is much more similar to that of T. keyi than to that of the other species of Echinosaura (see Fig. 4). There is a general north south trend in respect of relative snout length. The Panamanian and Colombian species (i.e., E. orcesi, E. palmeri, and E. panamensis) have markedly longer snouts than E. brachycephala, E. horrida, and T. keyi from Ecuador. However, in dorsal scalation E. brachy­cephala resembles E. palmeri and E. panamensis more than the geographically closer E. horrida (see Fig. 5). We strongly agree with Fritts et al. (2002) that palmeri and panamensis should be treated as separate species rather than subspecies of E. horrida as did Uzzell (1965a). The strikingly different tail morphology of T. keyi is obviously an autapomorphic character. Further studies are needed to evaluate the relationships of the species currently placed in Echinosaura and Teuchocercus. Echinosaura horrida and T. keyi have broadly overlapping geographic ranges (Fig. 6). In the vicinity of Mataje (330 m elevation, Esmeraldas Province), three of the species in question (E. orcesi, E. horrida, and T. keyi) appear to occur sympatrically, and at Paramba (770 m elevation, Imbabura Province), both E. horrida and T. keyi have been collected (Fritts et al., 2002; also data from specimens examined by authors). Echino­saura brachycephala, however, is currently known only from the foothills of the Andes and appears to have an allopatric distribution relative to T. keyi and the other species of Echinosaura. The documented elevational ranges of the species of Echinosaura and Teuchocercus are E. brachycephala (1275-1665 m), E. orcesi (250-820 m), E. horrida (200-860 m), E. palmeri (30-1520 m), E. panamen­sis (560-909 m), and T. keyi (280-860 m) (Fritts et al., 2002; also data from specimens examined by authors). Thus, E. brachycephala occurs at higher elevations as E. horrida and T. keyi. Key to the Species of Echinosaura and Teuchocercus la. Four ventral scales per caudal segment; postmental scale reduced or absent, not reaching beyond one-third of first infrala­bial ........................................Echinosaura brachycephala Fig. 5. Arrangement of tubercles on dorsum and proximal tail. (A) Echinosaura panamensis (ZFMK 49107); (B) Echinosaura palmeri (BMNH 1923.10.12.14); (C) Echinosaura horrida (ZFMK 43768); (D) Echinosaura brachycephala (ZFMK 46370); (E) Teuchocercus keyi (MFING 2284.59); (F) Echinosaura orcesi (NMW F 3377). Drawings by Mathias Gunther.60 G. KOHLER ET AL. lb. Three or 5-6 ventral scales per caudal segment; postmental scale not reduced, reaching beyond one-third of first infralabial . . 2 2a. Internasal single.............................................................3 2b. Internasal divided..........................................................4 3a. Three ventral scales per caudal segment; continuous paravertebral rows of enlarged scales present; 8-9 femoral pores (one side) in males, one in females...........Echinosaura horrida 3b. 5-6 ventral scales per caudal segment; continuous paravertebral rows of enlarged scales absent; 14-15 femoral pores (one side) in males, 5-6 in females .............................................................Echinosaura orcesi 4a. Some or all of tympanum covered by granular scales; tail of adults bearing whorls of greatly enlarged conical spines; postmen­tal scale small, not reaching beyond two- thirds of first infralabial................Teuchocercus keyi 4b. Some or all of tympanum not covered by granular scales; tail of adults not bearing whorls of greatly enlarged conical spines; postmental scale large, reaching beyond two-thirds of first infralabial.....................................5 5a. Frontal single................................Echinosaura palmeri 5b. Frontal transversely divided ..................................................Echinosaura panamensis Acknowledgments.-For the loan of or access to specimens we thank L. Ford and D. R. Frost, American Museum of Natural History (AMNH), New York; C. J. McCarthy, The Natural History Museum (BMNH), London; J. Hanken and J. P. Rosado, Museum of Comparative Zoology, Harvard University (MCZ), Cambridge; and J. Mariaux, Museum d'Histoire Naturelle (MHNG), Geneva; F. Tiedemann, Naturhistor- isches Museum Wien (NMW), Wien; C. A. Phillips and S. D. Sroka, Museum of Natural History, University of Illinois at Urbana (UIMNH), Urbana. We are grateful to T. H. Fritts, Fort Collins, who made available to us a prepublication copy of his and coworkers paper describing a new species of Echinosaura. T. H. Fritts and G. Smith, Granville, reviewed an early draft of our manuscript and made valu­able comments. We thank M. Gunther who contributed drawings and who shared unpub­lished data of his Echinosaura study with us. Literature Cited Breder Jr., C. M. 1946. Amphibians and reptiles of the Rio Chucunaque drainage, Darien, Panama, with notes on their life histories and habits. Bulletin of the American Museum of Natural History 86:375-436. Dunn, E. R. 1944. Notes in Colombian herpetology. II. The lizard genus Echinosaura (Teiidae) in Colombia. Caldasia 2:397-398. Fritts, T. H., and H. M. Smith. 1969. A new teiid lizard genus from western Ecuador. Transactions of the Kansas Academy of Science 72:54-59. Fritts, T. H., A. Almendariz, and S. Samec. 2002. A new species of Echinosaura (Gymnophthalmidae) from Ecuador and Colombia with comments on other members of the genus and Teuchocercus keyi. Journal of Herpetology 36:349-355. Leviton, A. E., R. H. Gibbs Jr., E. Heal, and C. E. Dawson. 1985. Standards in herpetology and ichthyology. Part I. Standard symbolic codes for institutional resource collections in herpetology and ichthyology. Copeia 1985:802-832. Uzzell, T. M. 1965a. Teiid lizards of the genus Echinosaura. Copeia 1965:82-89. ----------. 1965b. Teiid lizards of the genus Neusticurus (Reptilia, Sauria). Bulletin of the American Museum of Natural History 132:279-327. Accepted: 5 November 2003. Appendix 1 Comparative material examined (* = specimens used in the discriminant function analysis). Echinosaura orcesi.-Eeuador-Carchi: San Marco, 670 m: NMW 32000:1*; San Marco, 700 m: NMW 32000:2*. Echinosaura horrida.-Eeuador-"Ecuador": ZFMK 7269*-7271*, 7272; "Pacific versant of Ecuador": ZFMK 43757*, 43758, 43759-63 (all*), 43764, 43765-72 (all*), 43773-74, 43775*-76*, 43777-81; Esmeraldas: near Mataje, 350 m: NMW 32001:1; Imbabura: Lita: SMF 11752; Paramba: AMNH 13402-03, MCZ 11172; Man- abi: El Carmen, W of Santo Domingo de los Colorados: ZFMK 42760, 46369*; Pichincha: Centro Cientifico Rio Palenque, 200 m, 47 km S Santo Domingo de los Colorados: AMNH U9827*-28*, MCZ 147175,149670*­71*, 156140, 156148, 156149*, 156150-51, 156152*, 156153, 156154*, 156155-56; 4-5 km ESE El Esfuerzo, approximately 320 m in boulder strewn stream valleys: MCZ 171866*; Rio Baba, 5-10 km SSW Santo Domingo de los Colorados: AMNH 110612. Echinosaura palmeri.-Colombia-Cauca: Quebrada Guagui, 0.5 km above Ri'o Patia (upper Saija drainage), 100-200 m: AMNH 109695-97 (all*); Choco: Noananoa, Rio San Juan, 30 m: BMNH 1923.10.12.14; Quebrada Taparal, lower Rio San Juan (about 7 km airline NE Palestina); AMNH 123710*-ll*; Valle: near Cisneros on Buenaventura-Dagua Road: AMNH 108994. Panama- Darien: no specific data: AMNH 49186*-87*, 49190, 49195*; Chalichiman's Creek: AMNH 49200*. Echinosaura panamensis.-Panama-Cocle: El Valle de Anton, stream on N side, 610 m: AMNH 71707*-08; El Cope: ZFMK 45779*, 49107*, 50084*-85*, 50462*, 50464, 52200*, 54631*, 64837*; Panama: 2.5 km N Agua Clara rain guage on Santa Rita lumber Rd (E of Colon): MCZ 133722; Cerro Campana: MCZ 127763*. Teuchocercus kei/i.-Eeuador-Carchi: Rio Sabalera, 620-630 m: NMW 31988*, 32001:6-8; Ri'o Sabalera, near Ojala, 630 m: NMW 32001:2-4; Ojala, 400 m: NMW 32001:5; Esmeraldas: 1 km W El Placer: USNM 196094*; Rio Mira, 620 m, NMW 32001:9-11; Pichincha: 4 km E Rio Baba bridge, 24 km S Santo Domingo de los Colorados: UIMNH 80452; Puerto Quito: MHNG 2284.059*.