ZOOTAXA ISSN (online edition)

Zootaxa 1394: 25 45 (2007) www.mapress.com/zootaxa/ Copyright 2007 Magnolia Press ISSN 1175-5326 (print edition) ZOOTAXA ISSN 1175-5334 (online edition) A new species of Dendrelaphis (Serpentes: Colubridae) from Southeast Asia GERNOT VOGEL 1 & JOHAN VAN ROOIJEN 2 1 Society for Southeast Asian Herpetology, Im Sand 3, D-69115 Heidelberg, Germany. E-mail: Gernot.Vogel@t-online.de 2 Da Costastraat 99, 2321 AM Leiden, The Netherlands. E-mail: j1.van.rooijen@hetnet.nl Abstract A new species of the colubrid genus Dendrelaphis Boulenger 1890 is described. Dendrelaphis kopsteini sp. nov. ranges from Thailand through Peninsular Malaysia and Singapore to Sumatra. A detailed statistical analysis of the differences between D. kopsteini sp. nov., D. formosus (Boie, 1827) and D. cyanochloris (Wall, 1921) is provided as the three species have been mixed up frequently in the literature. D. kopsteini sp. nov. differs from all other Dendrelaphis species by a brick red neck coloration. A neotype is designated and described for D. formosus and a lectotype is designated and described for D. cyanochloris. Key words: Serpentes: Colubridae: Dendrelaphis: Dendrelaphis kopsteini sp. nov., Dendrelaphis cyanochloris; Dendrelaphis formosus, Dendrelaphis humayuni, Southeast Asia Introduction The colubrid snakes of the genus Dendrelaphis Boulenger 1890 are widely distributed, ranging from Pakistan in the West to the northern and eastern coast of Australia in the East and South and to southern China in the North (Ziegler & Vogel, 1999). Members of the genus Dendrelaphis are slender, diurnal species that are predominantly arboreal and feed on lizards and amphibians. The recent description of D. hollinrakei Lazell, 2002 has brought the number of known species to 21. Two Southeast Asian species, D. formosus (Boie, 1827) and D. cyanochloris (Wall, 1921), have been the subject of a long history of confusion. Starting in 1930, Smith identified a specimen as D. formosus but later rectified his identification to D. cyanochloris (Smith 1943). More than three decades later, Frith (1977) elaborated extensively on the problems he encountered with the differentiation between the two species. Nowadays, the confusion still persists. For example, Manthey & Grossmann (1997) and Lim & Lee (1989) portrayed under the name D. formosus a species clearly different from Stuebing & Inger s (1999) and Cox et al. (1997) taxon. One of these species is presented as D. cyanochloris in Chan-Ard et al. (2001). One reason for the existing confusion lies in the inadequacy of the original descriptions of D. formosus and D. cyanochloris. These descriptions lack a sufficient level of detail to allow for unequivocal differentiation between the two species. In addition, the original description of D. cyanochloris is composite as it was partly based on specimens of D. humayuni Tiwari & Biswas, 1973. Another reason for the confusion is the existence of a third, undescribed, species that is similar to D. formosus and D. cyanochloris in its pholidosis and, in most cases, has been identified as D. formosus in the literature. In this article, this new species of Dendrelaphis is described. In addition, descriptions of the types of D. formosus and D. cyanochloris are provided. Accepted by P. David: 29 Nov. 2006; published: 18 Jan. 2007 25

Materials and methods Given the similarity and consequent history of confusion of D. formosus, D. cyanochloris and D. kopsteini sp. nov., the true identities of D. formosus and D. cyanochloris were first determined by examination of the type specimens. Subsequently, a detailed statistical analysis was carried out with regard to the differences between D. formosus, D. cyanochloris and D. kopsteini. Finally, D. kopsteini was compared with other Southeast Asian members of the genus. For the statistical analyses, 33 D. formosus, 10 D. cyanochloris, 3 D. cf. cyanochloris and 8 D. kopsteini were examined. For each specimen, 21 characters bearing on colour pattern, body proportions and scalation were recorded. These are listed in Table 1. All colour pattern characters were coded as polarized (1,0) variables. Eye-diameter and distance eye-nostril were measured with a slide calliper to the nearest 0.1 mm. These measurements were made on the left and right side and were subsequently averaged. Distance from centre of the eye to posterior border of the nostril was determined by measuring the distance from the posterior margin of the eye to the posterior margin of the nostril and subtracting the radius of the eye. Snout-vent length and tail-length were measured by marking the length on a piece of string and subsequently measuring the position of the mark to the nearest 0.5 cm. Snout-vent length was measured to the posterior margin of the anal plate. The number of ventrals was counted according to Dowling (1951). Subcaudals were counted on one side, excluding the terminal scute. The first sublabial was defined as the scale that starts between the posterior chin shield and the infralabials and that borders the infralabials (see Peters, 1964). The last infralabial was defined as the last infralabial still covered completely by the last supralabial. Differences in growth stage distributions of the different samples may affect the analyses of morphometric variables. Therefore, EYED, EYEN and TAIL were regressed to a common SVL for each species separately using bivariate regression with SVL as independent (e.g. Thorpe, 1975, 1983; How et al., 1996; Turan, 1999). Homoscedasticity and linearity were verified by visually inspecting the plot of residuals versus predicted values. Normality of residuals was verified by Kolmogorov-Smirnov tests. The adequacy of the procedure was assessed by testing the significance of the correlation between the adjusted variables and SVL (e.g. Turan, 1999). Most univariate and multivariate analyses require an a priori grouping of specimens. As a consequence, additional groups of potentially taxonomic significance may go unnoticed. This is a disadvantage in the case of uncertain taxonomic status of one or more of the populations under study. The three specimens identified as D. cf. cyanochloris represented such a population from Java. PCA is a multivariate technique that does not require a priori grouping (e.g. Turan, 1999; Wüster et al., 2001). Therefore, before further analysis, a PCA was performed to check the grouping of the specimens into three groups, i.e. D. cyanochloris, D. formosus and D. kopsteini. For this purpose, a preliminary univariate analysis was run to select variables that showed strong taxonomic relevance. This selection was done to ensure loading of relevant variables on the first two principal components. The grouping was subsequently checked by visual inspection of the plot of the first two principal components. Subsequent analyses of the differences between the three species were first carried out univariately. Quantitative variables were log-transformed before analysis and were analysed using one-way ANOVA unless the assumptions underlying this technique were violated. The normality-assumption was tested using the Kolmogorov-Smirnov test and the homogeneity of variance was tested using Levene s test. In cases where the assumptions were violated, the nonparametric Mann Whitney U test was used. Qualitative (polarized) variables were analysed using Chi-square. Finally, a discriminant analysis was applied to the morphological characters to determine multivariate significance levels and to visualize the separation of the species in multivariate space. D. kopsteini was compared with D. humayuni primarily on the basis of literature as these species can not be confused. However, EYED and TAIL could not be compared directly for reasons stated above. Therefore, 26 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

individual data on EYED, TAIL and SVL of D. humayuni were taken from Tiwari & Biswas (1973) and from one specimen examined by us. Subsequently, EYED and TAIL of D. humayuni and D. kopsteini were adjusted to a common SVL and compared univariately according to the methods described above. This comparison is treated separately from the comparsion of D. kopsteini, D. cyanochloris and D. formosus. TABLE 1. List of morphometric, meristic and coloration characters used in this study and their abbreviations. Abbreviation Morphometrics EYED EYEN TAIL SVL Scalation VENT SUBC SUBL SPL1 SPL2 POC INFR TEMP DOR1 DOR2 DOR3 LOR Coloration STR1 STR2 TSTR VERT TIP Character Horizontal diameter of the eye Distance from centre of the eye to posterior border of the nostril Tail-length Snout-vent length Number of ventrals Number of subcaudals Number of infralabials touched by the first sublabial Number of supralabials Number of supralabials touching the eyes Number of postoculars Number of infralabials Number of temporals Number of dorsals 1 head-length behind the head Number of dorsals at the position of the middle ventral Number of dorsals 1 head-length before the tail Number of loreals Presence or absence of three lateral black stripes on the posterior half of the body Presence or absence of a light ventrolateral stripe Postocular stripe covers complete temporal region and runs onto the neck or covers only the lower half of the temporal region and ends at the rear of the jaw Vertebral scales with or without broad black posterior margin Subcaudals with or without black tip Variables that were taxonomically most important were subjected to ANOVA to determine whether any of these variables showed sexual dimorphism. If so, a two-way ANOVA (species x sex) was performed to determine whether the intersexual differences were relevant in the context of interspecific differences. All statistical analyses were carried out with the software SPSS (2003; SPSS for Windows. Release 11.5.2.1. Standard Version. SPSS Inc., Chicago). Museum abbreviations: BMNH: Natural History Museum, London, Great Britain. MNHN: Muséum National d Histoire Naturelle, Paris, France. PSGV: Gernot Vogel s private collection, Heidelberg, Germany. QSMI: Queen Saovabha Memorial Institute, Thai Red Cross Society, Bangkok, Thailand. RMNH: National Museum of Natural History, Leiden, The Netherlands. SMF: Natur-Museum und Forschungs-Institut Senckenberg, Frankfurt-am-Main, Germany. ZMA: Zoological Museum Amsterdam, The Netherlands. ZMB: Zoologisches Museum für Naturkunde der Humboldt-Universität zu Berlin, Berlin, Germany. ZSM: Zoologische Staatssammlung, München, Germany. A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 27

Results Statistical analyses Morphometric variables (EYED, EYEN, TAIL) were regressed to a common SVL of 73.5 cm. The adjusted variables showed no significant correlation with SVL, therefore the size effect had been eliminated. A preliminary univariate analysis demonstrated that EYED, VENT, SUBC, SUBL, TSTR, STR1, VERT and TIP were taxonomically important variables. These were used in a categorical PCA to check the existence of three groups (D. cyanochloris, D. formosus and D. kopsteini). This checking was applied as there was doubt with regard to the taxonomic status of three specimens from Java. These were putatively identified as D. cf. cyanochloris although originally identified as D. formosus according to the museum labels. D. cyanochloris has never been recorded from Java and the three specimens would represent an insular population located approximately 2000 km from the southernmost locality of the mainland population of D. cyanochloris. Figure 1 shows a plot of the first two PC s. The three specimens identified as D. cf. cyanochloris appear to constitute a separate cluster and therefore might represent an independent insular lineage. Consequently, these specimens were excluded from further analysis and will be subjected to future additional research (Van Rooijen & Vogel, in prep.). Furthermore, the plot shows a clear separation between D. cyanochloris, D. formosus and D. kopsteini. After exclusion of these specimens, morphometric variables (EYED, EYEN, TAIL) were regressed to a new common SVL of 74.5 cm and effectiveness of the procedure was checked again. FIGURE 1. Results of a principal component analysis based on taxonomically relevant morphometric, meristic and coloration characters. Scores on the first two principal components have been plotted to check the a priori grouping into three species. Three specimens from Java, putatively identified as D. cf. cyanochloris, appear to constitute a separate group and are consequently excluded from further analysis. 28 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

TABLE 2. Descriptive statistics with regard to the examined characters. Mean and range are shown in case of continuous quantitative variables (EYED-TAIL). Median and range are shown in case of discrete quantitative variables (VENT- LOR). In case of qualitative variables (STR1-TIP), the percentage of specimens possessing the indicated character is given. EYED, EYEN and TAL represent the SVL-adjusted values. The maximum total length (TLMAX) is presented in the last row. TLMAX was not tested statistically for obvious reasons. Significance-levels of differences between the three species are given in the last three columns. D. kopsteini (n=8) D. formosus (n=33) D. cyanochloris (n=10) kopsteini / formosus kopsteini / cyanochloris formosus / cyanochloris EYED (mm) 5.9 (5.5 6.2) 6.3 (5.8 6.9) 5.3 (4.7 5.7) 0.0005 0.001 <0.00001 EYEN (mm) 7.7 (7.3 7.9) 7.9 (6.9 8.7) 7.4 (6.8 8.0) - - 0.01 TAIL (cm) 40.0 (38.5 41.0) 37.0 (32.5 43.5) 34.0 (32.0 38.0) 0.003 0.0001 0.001 VENT 176 (167 181) 181 (172 194) 200 (187 205) 0.001 <0.00001 <0.00001 SUBC 147 (140 154) 149 (142 162) 142 (137 156) - - 0.004 SUBL 4 (4 4) 8 (5 10) 4 (4 9) <0.00001-0.002 SPL1 17 (16 19) 18 (17 18) 18 (16 18) 0.03 - - SPL2 4 (4 4) 6 (4 6) 6 (4 6) 0.001 0.02 - POC 4 (4 4) 4 (4 9) 4 (4 4) - - 0.05 INFR 20 (20 21) 20 (19 22) 20 (18 22) - - - TEMP 12 (11 14) 11 (8 14) 12 (8 12) - - - DOR1 15 (15 15) 15 (15 15) 15 (15 16) - - - DOR2 15 (15 15) 15 (13 15) 15 (15 15) - - - DOR3 11 (11 11) 11 (11 13) 11 (11 11) - - - LOR 1 (1 1) 1 (1 1) 1 (1 1) - - - STR1 (%) 0 100 0 <0.00001 - <0.00001 STR2 (%) 0 0 0 - - - TSTR (%) 0 100 100 <0.00001 0.00002 - VERT (%) 100 0 0 <0.00001 0.00002 - TIP (%) 100 72 0-0.00002 0.0001 TLMAX (cm) 143 147 143 - - - Descriptive statistics and univariate comparisons are summarized in table 2. Figure 2 shows the results of a discriminant analysis applied solely to the morphological characters. Characters that showed no significant difference in univariate analysis were excluded. The three species clearly constitute morphologically distinct clusters. Both discriminant functions were highly significant (p<0.000001) based on Wilk s lambda. Due to the fact that some variables violated the assumptions of discriminant analysis these levels should be viewed with some reserve. However, given the very high significance-levels and the number of significant univariate differences, there is no doubt about the validity of the separation between the species. Based on the correlations between the original variables and the discriminant functions, the variables EYED, VENT and SUBL were the most discriminating variables. One-way ANOVA s showed a significant sexual dimorphism in EYED in D. formosus (a mean of 6.45 mm in males versus 6.25 mm in females; SVL-adjusted values; p=0.025, see also fig. 16). In that regard it is interesting to note that the number of postoculars is also sexually dimorphic in D. formosus, with males possessing more postoculars on average (5.3 in males versus 4.3 in females; p=0.03; Mann-Whitney U test). This suggests a relation between eye-size and number of postoculars, with a greater eye-size being accompanied by a higher number of postoculars. Sexual dimorphism in EYED was not established in the other species. More specimens of D. formosus were examined than specimens of D. cyanochloris and D. kopsteini. Consequently, A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 29

the statistical power was higher in analyses with regard to D. formosus. Therefore, similar sexual dimorphism in eye-size may occur in the other species as well, but could not be established. A two-way ANOVA with species and sex as factors demonstrated no significant effect of sex on eye-size whereas the effect of species on eye-size was highly significant (p<0.000001). Therefore, the sexual differences in eye-size are negligible in comparison to the interspecific differences. A significant sexual dimorphism with regard to VENT and SUBL was established in none of the species. These results are consistent with findings by How et al. (1996) with regard to (former) subspecies of D. pictus. How et al. (1996) found no sexual dimorphism in VENT but did find sexual dimorphism in EYED. In addition, in the context of interspecific differences, the sexual difference in EYED was negligible. FIGURE 2. Results of a discriminant analysis based on morphological characteristics. Scores on the first two discriminant functions have been plotted to illustrate the separation of Dendrelaphis kopsteini sp. nov., Dendrelaphis formosus and Dendrelaphis cyanochloris in multivariate space. The three species clearly constitute morphologically distinct clusters. Descriptions of type-specimens Designation and description of a neotype of Dendrelaphis formosus As was usual at that time, Friedrich Boie did not designate a type specimen in his description of Dendrophis formosa. Nevertheless he gave a rather complete description of one specimen. At that time Boie was working in the Museum of Leiden. So the specimen mentioned by him should have been deposited in that collection. There is one specimen of this species in Leiden that was already present in 1827 (RMNH 877). It was collected by H. Boie and H. Macklot on Java. The original notes are not readable. We compared this specimen with Boie s description. Unfortunately the counts of the ventrals and subcaudals differ from this specimen as follows: Boie counted 185 ventrals and 145 subcaudals, we counted 178 ventrals (according to Dowling) and 30 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

143 subcaudals. Even if there are differences in the way of counting these scales, they cannot explain this large difference especially in the ventral counts. However, it is well known that in these former times the scales were not counted very accurately, so we believe this animal to be the specimen Boie had to hand since apart from these counts the description matches perfectly. On the basis of this historical indication, we designate RMNH 877 as the neotype of Dendrophis formosa Boie, 1827. As the description is in Latin language, we here give a translation: With large eyes, medium tail, loreal scale narrowing backwards [occipitalibus pone obtusis], 2 (left) temporal scales. Completely blue, the underside paler, a stripe runs through the eye, and some lateral black stripes posteriorly on the body, all the scales on the side of the body and on the tail are black-edged. 185 [ventrals] 145 [subcaudals]. Native in Java. A complete description of the neotype RMNH 877 follows: FIGURE 3. Dendrelaphis formosus. Neotype RMNH 877. Photograph by Johan van Rooijen. FIGURE 4. Dendrelaphis formosus. Neotype RMNH 877. Left side of the head. Photograph by Johan van Rooijen. Adult specimen from Java; SVL 99.0 cm; TAIL 48.0 cm; Head-length 32.7 mm; 178 ventrals; 143 divided subcaudals; anal plate divided; dorsals 15:15:11; vertebral scales larger than first row of dorsals, hexagonal in shape and with clearly concave posterior margin; 1 loreal (R); loreal fused with prefrontal (L); 1 preocular (L+R); 3 postoculars (L+R); 1 supraocular (L+R); 9 supralabials (L+R); supralabials 4,5 and 6 touch the eye (4 th with posterior corner) (L+R); 10 infralabials, 6 th the largest (L+R); first infralabials touch at the mental groove; infralabials 1 through 5 touch anterior chin shield (L+R); infralabials 5 and 6 touch posterior chin shield (L+R); first sublabial very long, touching infralabials 6 through 9 (L+R); temporals 2:2:2 (L+R); eye very large, 8.0 mm in diameter (L and R averaged); the head bulges somewhat due to the accommodation of the large eye; ground-colour blue; a black temporal stripe starts on the rostral scale, covers the lower half of the nasals, the whole loreal, the lower half of the preocular, the whole temporal region and extends onto the A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 31

neck up to the level of the 12 th ventral scale; all dorsals in the anterior half of the body, except first row, have a white spot on the lower edge, most or all of this spot is covered by lower adjoining dorsal scales; all dorsals have black anterior and lower edges which are largely covered by preceding and adjoining dorsals; the vertebral scales have black edges anteriorly which are largely covered by preceding vertebral scales and adjoining dorsals; all vertebral scales have white lateral margins which are largely covered by adjoining dorsals; subcaudals have a black point medially; 3 pronounced lateral black stripes in posterior half of the body; the first black stripe is formed by black edges of the ventrals and first dorsals, the second and third are each formed by a black upper edge and a black lower edge of two adjoining dorsal rows; the dorsals on the tail have a pronounced black circumference resulting in a reticular black pattern on the tail; no light ventrolateral stripe (fig. 3 6). FIGURE 5. Dendrelaphis formosus. Living specimen from Bako, Borneo. Photograph by Björn Lardner. FIGURE 6. Dendrelaphis formosus. Living specimen from Santubong Peninsula, Sarawak, Borneo. Photograph by Johan van Rooijen. 32 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

Designation and description of a lectotype of Dendrelaphis cyanochloris In his description of Dendrophis pictus variety B cyanochloris, Wall (1921) listed several specimens that he had to hand. These are 19 specimens from the Indian Museum and 2 specimens from Bombay collection. They originated from Eastern Himalayas, Assam, Burma and Nicobars. Obviously two different species were included in this series. The specimens from the Nicobar Islands should be regarded as Dendrelaphis humayuni Tiwari & Biswas, 1973. None of the listed specimens is described in detail or designated as holotype. Consequently, all have to be regarded as syntypes. Later Wall gave several specimens to the British Museum. He indicated one as holotype (BMNH 1946.1.6.13, McCarthy pers. comm.) but, as this was not published, it has no nomenclatural standing. Unfortunately this specimen was not included in the type series as can be seen by its locality Mergui, Burma. So it cannot be construed as a potential holotype or lectotype. Consequently, we take this opportunity to designate BMNH 1940.3.4.27 as lectotype. This specimen originates from Darjeeling and was transferred from the collection of the Indian Museum. It has the original collection number attached on a tag as 7734. It was included in the syntype series. Thus, Darjeeling becomes the valid type locality for Dendrelaphis cyanochloris. A complete description of the lectotype BMNH 1940.3.4.27 follows. FIGURE 7. Dendrelaphis cyanochloris. Lectotype BMNH 1940.3.4.27. Photograph by Johan van Rooijen. FIGURE 8. Dendrelaphis cyanochloris. Lectotype BMNH 1940.3.4.27. Left side of the head. Photograph by Gernot Vogel. Subadult male from Darjeeling, State of West Bengal, India.; SVL 59.0 cm; TAIL 26.0 cm; Head-length 19.0 mm; 201 ventrals; 137 divided subcaudals; anal plate divided; dorsals 15:15:11; vertebral scales larger than first row of dorsals, hexagonal in shape and with straight posterior margin; 1 loreal (L+R); 1 preocular (L+R); 2 postoculars (L+R); 1 supraocular (L+R); 9 supralabials (L+R); supralabials 5 and 6 touch the eye A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 33

(L+R); 9 infralabials, 5 th largest (L), 10 infralabials, 6 th largest (R); first infralabials touch at the mental groove; infralabials 1 through 4 touch anterior chin shield (L), infralabials 1 through 5 touch anterior chin shield (R); infralabials 4 and 5 touch posterior chin shield (L); infralabials 5 and 6 touch posterior chin shield (R); first sublabial long, touches infralabials 5 through 9 (L), first sublabial short, touches infralabials 6 and 7 (R); temporals 2:2:2 (L+R); eye small, 4.3 mm in diameter (L and R averaged); a black temporal stripe starts behind the eye, covers the whole temporal region and extends onto the neck up to the level of the 4 th ventral scale. In the anterior part of the body, there are groups of dorsals with white spots, these white spots are partly covered by adjoining dorsals; all dorsals, except first row, have black anterior and lower edges which are largely covered by preceding and adjoining dorsals; vertebral scales with black edges anteriorly which are largely covered by preceding vertebrals and adjoining dorsals; some vertebral scales lack a white lateral margin, some have a unilateral white margin and some have white margins bilaterally, these white margins are covered by adjoining dorsal scales; subcaudals without a black point medially; no lateral black stripes in posterior half of the body; dorsals on tail without black margins; no light ventrolateral stripe (fig. 7 8) Dendrelaphis kopsteini sp. nov. (Figs. 9 14) Ahaetulla formosa (non Dendrophis formosus Boie, 1827): Taylor (1965: 814). Dendrelaphis formosus: Flower (1896: 883); Tweedy (1983: 63, 154, plate 4); Lim & Lee (1989: 53); Vogel (1990: 10, Abb. 4); Lim & Lim (1992 : 64) ; Manthey & Grossman (1997: 337, Abb. 246); Ziegler & Vogel (1999: 206); Pauwels et al. (2000: 141); Nutphand (2001: 148). Dendrophis formosa: Frith (1977: 278)? Dendrelaphis formosus (non Dendrophis formosus Boie, 1827) part.: Flower (1899: 605, 660); Inger & Voris (2001: 889); Iskandar & Colijn (2002: 53). Dendrelaphis spec. A: Rubeli (1988: 133.) This chresonymy includes only citations based on specimens definitely identified as Dendrelaphis kopsteini sp. nov. Holotype. MNHN 1962.1052, adult female (figs. 9 12), from fleuve Endau, état de Johore, Endau river, State of Johore, Federation of Malaysia. Collected by Pierre Pfeffer, unknown date. Paratypes (7 specimens). Thailand. MNHN 1997.6579 (female) Phang-Nga Province, Wildlife breeding station, Muang District ; QSMI Q 531 (former PSGV 45) (female), near Thung Song, Nakhon Si Tammarat Province, South Thailand. Federation of Malaysia (West Malaysia). MNHN 1962.1051 (male), fleuve Endau, état de Johore meaning Endau river, Province of Johore, Federal Republic of Malaysia; ZMH R08445 (former PSGV 269) (male), way to Cameron Highlands, 620m, on the highway to Cameron Highlands, Perak province, 620m; RMNH 40117 (former PSGV 777) (male), Pet Trade, West Malaysia. Singapore. MNHN 1893.147 (male), Singapore. INDONESIA. Sumatra Island. ZSM 277/1909 (female), Tanjung Morawa, located near Medan, Sumatera Utara Province. Diagnosis. A rather stout species of the Dendrelaphis formosus group, with 15 dorsal scale rows at midbody, 167-181 ventral scales, 140-154 subcaudal scales, 2 supralabials touching the eye and a first sublabial that touches 2 infralabials. The eye is rather large. The vertebral scales are larger than the lowest dorsal row. There is one loreal plate. A black postocular stripe covers only the lower half of the temporal region and ends at the rear of the jaw. Vertebral scales with a broad black posterior margin. It differs from all other species of this genus by the red colour that is visible in life on the skin between the scales of the first part of the body. Etymology. This species is named after Dr. Felix Kopstein (1893-1939), who contributed so much to our knowledge of the snake fauna of Indonesia. Several of the specimens we used for our studies were collected by him. He was also the first to report delayed fertilisation. Suggested English name: Kopstein s Bronzeback. 34 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

Description of the holotype (figs. 9-12). Body slender; head long, distinct from the neck, rather broad for a species of this genus, when seen from the side. Eye rather large, 7.1 mm in diameter. Pupil round. Tail long. The reddish tongue is protruded. SVL 95.0 cm; TAIL 46.0 cm; TL 141.0 cm; ratio TAIL/TL: 0.33; 181 ventrals (+3 preventrals); 147 divided subcaudals (plus 1 terminal scale); anal shield divided; dorsals 15:15:11; dorsal scales disposed obliquely, smooth and without apical pits; vertebral scales enlarged, larger than first row of dorsals, hexagonal in shape and with straight posterior margins; 1 loreal (L+R); 1 preocular (L+R); 2 postoculars (L+R); 1 supraocular (L+R); 8 supralabials (L+R); supralabials 4 and 5 touch the eye (L+R); 10 infralabials (L), 11 infralabials (R); first infralabials touch at the mental groove; infralabials 1 through 5 touch anterior chin shield (L+R); infralabials 5 and 6 touch posterior chinshield (L+R); first sublabial short, touches infralabials 6 and 7 (L+R); temporals 2:2:2 (L+R). FIGURE 9. Dendrelaphis kopsteini sp. nov. Holotype, MNHN 1962.1052. Left side of the head. Photograph by Gernot Vogel. Ground-colour bronze; a black temporal stripe starts before the eye, covers the lower half of the temporal region and ends at the rear of the jaw. The throat is yellowish with the sublabials being partly bluish. The vertebral scales have a very broad black posterior edge. Some vertebral scales lack a white lateral margin, some have a unilateral white margin and some have white margins bilaterally; subcaudals have a black point medially; no lateral black stripes in posterior half of the body; dorsals on the tail have a pronounced black circumference resulting in a reticular black pattern on the tail; no light ventrolateral stripe. Belly dark. Description of the paratypes. A summary of morphological and coloration data of the paratypes is given in Table 3. Other important characters agree with features of the holotype. A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 35

TABLE 3. Morphological and coloration characters of the paratypes of Dendrelaphis kopsteini. MNHN 1962.1051 MNHN 1893.147 ZMH R08445 (former PSGV 269) Collection number RMNH 40117 (former PSGV 777) MNHN 1997.6579 QSMI Q 531 (former PSGV 45) ZSM 277/1909 Sex M M M M F F F SVL (cm) 78.5 96.5 75.5 81.5 88.0 91.0 81.5 Postoculars 2/2 2/2 2/2 2/2 2/2 2/2 2/2 TaL (cm) 39.5 46.0 41.5 41.5 43.5-43.0 Ventrals 175 179 171 173 177 178 167 Subcaudals 154 149 141 148 140-146 Dorsal formula 15:15:11 15:15:11 15:15:11 15:15:11 15:15:11 15:15:11 15:15:11 Infralabials 10/10 10/10 10/10 10/10 10/10 10/10 10/10 Supralabials 9/10 9/9 9/? 8/9 9/8 8/9 8/8 Supralabials touching the eye 5,6/5,6 5,6/5,6 4,5/4,5 4,5/5,6 5,6/4,5 4,5/5,6 4,5/4,5 Infralabials touched by the first sublabial 6,7/6,7 6,7/6,7 -/6,7 6,7/6,7 6,7/6,7 6,7/6,7 6,7/6,7 Temporal formula 2:2:2/2:2:2 2:1:2:2/2:2:1:2 2:2:2/- 2:2:2/2:2:2 2:1:2/2:2:2 2:2:2/2:2:2 2:2:2/2:2:2 Eye-diameter (mm) 6.4/6.4 6.9/7.0 -\- 5.8/5.9 6.4/6.5 6.5/6.5 6.3/6.4 Vertebral scales larger than first row of dorsals yes yes yes yes yes yes yes Temporal stripe covers lower part of temporal yes yes yes yes yes yes yes region and ends at rear of jaw Vertebral scales with broad black posterior margin yes yes yes yes yes yes yes Subcaudals with black tip yes yes yes yes yes yes yes White ventrolateral stripe present no no no no no no no 36 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

FIGURE 10. Dendrelaphis kopsteini sp. nov. Holotype, MNHN 1962.1052. Top of the head. Photograph by Gernot Vogel. Description and variation. The maximal total length known is 142.5 cm (SVL 96.5 cm, TAIL 46.0 cm) for a male (MNHN 1893.147 from Singapore). The largest known female is 131.5 cm long (SVL 88.0 cm, TAIL 43.5 cm; MNHN 1997.6579 from Phang-Nga, Thailand). This species reaches a size similar to D. cyanochloris and D. formosus. Body relatively slender in males and females; head long, distinct from the neck, rather broad for a species of this genus, when seen from the side. Eye rather large, 5.8 7.1 mm in diameter (unadjusted values). Diameter eye / distance (anterior border of) eye to nostril 1.1 1.4 in 3 males, 1.2 1.3 in 4 females. Pupil round. Relative tail length 0.32 0.34. No sexual dimorphism was noted for the relative tail length. Dorsals smooth, in 15:15:11 rows; 171 179 ventrals (plus 1 3 preventrals) in 4 males, 167 181 ventrals (plus 1 3 preventrals) in 4 females; 141 154 subcaudals in 4 males, 140 147 subcaudals in 3 females, all paired; anal shield divided; 1 preocular; 2 postoculars; 1 loreal; normally 2+2+2 temporals (2+1+2+2, 2+2+1+2, 2+1+2 each in one case); 8-10 supralabials (9/10 in 1 specimen, 9/9 in 1 specimen, 9/8 in 3 specimens, 8/8 in 2 specimens, one damaged specimen has 9 on one side); in all cases 2 supralabials touching the eye (4/5 on both sides in 3 specimens, 5/6 on both sides in 2 specimens, 4/5 on one side and 5/6 on the other side in 3 specimens); 10 infralabials, one specimen has 11 infralabials on one side; 2 pairs of gulars between the preventrals and the posterior submaxillaries; 6 th and 7 th infralabial touched by the first sublabial. Ground colour bronze, olive or brown, the vertebral scales sometimes lighter with broad black posterior margins. The neck is brick red anterior followed by blue coloration. This coloration is extremely vivid when the snake inflates its neck. This vivid coloration reduces to some blue dots after the first third of the body. The A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 37

posterior part of the dorsal scales is dark, forming small vaguely defined bands. There are no black lateral stripes and there is no light ventrolateral stripe. The colour of the head corresponds to the ground coloration dorsally. A dark line extends from the nose to the angle of the mouth, covering only the lower temporal region. The upper labials are yellowish as is the underside of the head. The anterior part of the belly is yellowish, darkening towards the rear. The colour of the tail corresponds to the ground colour, dorsally the scales have dark margins. The subcaudals have dark tips forming some kind of a median line. The tongue is red, the eye brown. In alcohol the red colour vanishes very fast, the blue coloration vanishes after several years. We noted no sexual dimorphism in our sample of 4 males and 4 females. FIGURE 11. Dendrelaphis kopsteini sp. nov. Holotype, MNHN 1962.1052. Ventral side of the head. Photograph by Gernot Vogel. Range. The examined specimens of D. kopsteini originated from Peninsular Thailand, Peninsular Malaysia, Singapore and Sumatra. Comparison with other species. Evidently, there are many significant differences between D. kopsteini, D. cyanochloris and D. formosus (Table 2). The most prominent morphological differences are as follows. The eye is very large in D. formosus whereas it is small in D. cyanochloris (see also fig. 15). D. kopsteini takes up an intermediate position. Although this was not quantified and tested, the shape of the head corresponds with the eye-size. In D. formosus, the head bulges behind the snout to accomodate the large eye whereas the head is nearly flat in D. cyanochloris (fig. 15). Again, D. kopsteini takes up an intermediate position. The three species differ in tail-length which is highest in D. kopsteini. The three species differ significantly in ventral count which is highest in D. cyanochloris and lowest in D. kopsteini. D. formosus has a 38 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

higher subcaudal count than D. cyanochloris. A conspicuous difference lies in the length of the first sublabial. In D. formosus, this shield is very long whereas it is short in both D. kopsteini and D. cyanochloris. Finally, the eye is always touched by two supralabials (4,5 or 5,6) in D. kopsteini whereas in most cases it is touched by three supralabials (mostly 4,5,6 where 4 touches the eye with the posterior corner) in D. formosus and D. cyanochloris. The differences in coloration are pronounced and highly invariable. D. formosus always has three black lateral stripes in the posterior half of the body whereas these stripes never occur in D. cyanochloris or D. kopsteini. The black stripes in D. formosus are a prominent feature and are even explicitly mentioned in Boie s overly concise description. In D. kopsteini the postocular stripe covers only the lower half of the temporal region and ends at the rear of the jaw. In contrast, the postocular stripe covers the whole temporal region and runs onto the neck in both D. formosus and D. cyanochloris. D. kopsteini has vertebral scales with a very broad black posterior margin in contrast to D. formosus and D. cyanochloris. Finally, in both D. formosus and D. kopsteini the subcaudals often have black tips whereas this is never the case in D. cyanochloris. The differences in eye-size and tail-length between D. kopsteini and D. humayuni were determined separately (see materials and methods) and are not shown in table 2. D. kopsteini differs from D. humayuni in having a much larger eye (5.5-6.2 versus 4.0-5.4 (SVL-adjusted values); p<0.0001) and a longer tail (38.5 41.0 cm versus 32.5 39.0 cm (SVL-adjusted values); p=0.0002). In addition, the postocular stripe ends at the rear of the jaw in D. kopsteini. In D. humayuni, it runs onto the neck in where it breaks up into black spots (based on the description by Tiwari & Biswas, 1973; picture in Vijayakumar & David, 2006 and examination of BMNH 1940.3.4.28). Finally, Dendrelaphis kopsteini differs from all other members of the genus by its brick red nape, which makes it very easy to identify if it is alive or freshly dead. FIGURE 12. Dendrelaphis kopsteini sp. nov. Holotype, MNHN 1962.1052. Photograph by Gernot Vogel. A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 39

Apart from the differences, there are some important similarities. All species of this group have strongly enlarged vertebral scales that are larger than the first row of dorsals, 15 dorsals at mid-body and a single loreal scale. Dendrelaphis-species have either 15 or 13 dorsals at mid-body and Mertens (1934) considered the possession of 15 dorsals to be a primitive character state. Finally, all lack the light ventrolateral stripe which is present in several Dendrelaphis-species and was considered to be a primitive character state by Mertens (1934). Pending phylogenetic analysis of this genus, we define the D. formosus group on the basis of these similarities. The following species are members of this group: D. kopsteini, D. cyanochloris, D. formosus, D. humayuni, D. cf. cyanochloris (Java). FIGURE 13. Dendrelaphis kopsteini sp. nov. Living specimen from Thailand. Photograph by Gernot Vogel. FIGURE 14. Dendrelaphis kopsteini sp. nov. Living specimen from Thailand. Photograph by Gernot Vogel. Discussion. In order to evaluate the status of D. kopsteini, D. cyanochloris and D. formosus, the distributional ranges of the latter two species need to be revised. The examined specimens of D. cyanochloris originated from Northeast India, Burma and Thailand with the Thai Peninsula being the southernmost record. Another record from the Thai Peninsula, (Frith, 1977) is trustworthy given the morphological data provided (213 ventrals, 151 subcaudals). A record from Peninsular Malaysia (Chan-Ard et al., 2001) without a doubt is a D. formosus as the shown specimen has black lateral lines and a large eye. The records from Pulau Tioman, 40 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

West Malaysia (e.g. Hien et al., 2001; Van Rooijen & Van Rooijen, 2002) are questionable. The portrayed specimens appear to be D. formosus although they might even represent an undescribed taxon. Finally, Wall (1921) mentioned the Nicobar islands as part of the range but these records turned out to represent D. humayuni (Tiwari & Biswas, 1973; Vijayakumar & David, 2006). Indeed, a specimen from the Nicobars examined by us (BMNH 1940.3.4.28) also turned out to be a D. humayuni. The examined specimens of D. formosus originated from Peninsular Thailand, Peninsular Malaysia, Sumatra (including the Mentawei Islands), Java and Borneo. As such, all three species occur sympatrically in Peninsular Thailand. In addition, D. kopsteini occurs sympatrically with D. formosus in Peninsular Malaysia and Sumatra. FIGURE 15. Comparison of eye-size and head-shape between D. formosus (RMNH 40098, left) and D. cyanochloris (BMNH 1909.3.9.10, right), both subadult males. Notice the remarkably large eye in D. formosus and the fact that the head bulges behind the snout to accommodate the large eye. Photograph by Johan van Rooijen. FIGURE 16. Comparison of eye-size of a male (above) and a female (below) D. formosus. Photograph by Gernot Vogel. D. kopsteini, D. formosus and D. cyanochloris differ from each other with regard to coloration, scalation and body-proportions. The three species are discretely diagnosable on the basis of several characters and there is no indication of hybridization in areas of sympatry. As such, whether the Biological Species Concept or the Evolutionary Species Concept (e.g. Frost & Hillis, 1990) is applied, there is no doubt that these species represent independent lineages. An updated key to the Dendrelaphis-species of the Sunda-region is provided A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 41

below. D. humayuni is included as it is a member of the D. formosus group. It is entirely possible that more undescribed species inhabit the Sunda-region. Therefore, the key may not work in all cases. 1 13 dorsal scale rows at mid-body...caudolineatus - 15 dorsal scale rows at mid-body... 2 2 a light ventrolateral stripe present...pictus - no light ventrolateral stripe... 3 3 temporal stripe covers only the lower part of the temporal region... 4 - temporal stripe covers the whole temporal region... 5 4 temporal stripe ends at the rear of the jaw; neck red when inflated; eye large; 2 supralabials touch the eye; first sublabial touches 2 infralabials... kopsteini sp. nov. - the temporal stripe runs onto the neck where it breaks up into black bars; neck not red; eye small......humayuni 5 neck yellow when inflated; black oblique bars laterally on the body; less than 170 ventrals... striatus - neck not yellow, more than 170 ventrals... 6 6 3 black lateral stripes present on the posterior third of the body; eye very large, head bulges behind the snout to accomodate the large eye; first sublabial touches more than 2 infralabials...formosus - 3 black lateral stripes absent; eye small, head nearly flat, more than 180 ventrals......cyanochloris (mainland South-East Asia) / cf. cyanochloris (Java) Given the number of highly significant differences between D. kopsteini, D. formosus and D. cyanochloris, it is amazing that these species have been confused for such a long time. On the other hand, early confusions have led to mistakes in the literature which have led to subsequent misidentifications. Such vicious circles can only be broken by thorough re-examination of the original material. For example, Boulenger (1894) mentions a ventral range of 179-205 and a subcaudal range of 132-158 for D. formosus. These data have been copied in subsequent major works such as De Rooij (1917), Meise & Henning (1932) and Taylor (1965). The subcaudal minimum of 132 is very low whereas the ventral maximum of 205 is very high compared to our data (table 2). The source of the low subcaudal minimum mentioned by Boulenger (1894) was a specimen originating from Afghanistan (??). Examination of this specimen (BMNH 43.7.21.68) showed that it was indeed not a D. formosus, although it could not be identified. The high ventral maximum of 205 mentioned by Boulenger was based on a specimen from Java. The specimen was not examined in this study, but on the basis of the collected data on formosus-specimens from Java (n=20), the probability to observe a specimen with 205 ventrals was calculated to be lower than 0.00005. Given the low number of specimens examined by Boulenger (n=6), his extremely high maximum ventral count is highly unlikely. Furthermore, the above-mentioned probability was calculated on the basis of a theoretical normal distribution. As biological variables do not follow theoretical normal distributions to the extremes of the distribution, the count provided by Boulenger is probably non-existent in reality. As such, it is evident that the initial ranges in ventral and subcaudal counts as given by Boulenger were not based solely on D. formosus. This idea is strengthened by Boulenger s statement that black lateral lines may be present in the posterior part of the body of D. formosus. In fact, all specimens examined by us (n=33), from the whole distributional range, possessed black lateral stripes posteriorly. Similar shortcomings held true for the description of D. cyanochloris. Wall (1921) mentioned a ventral count of 175-207 in the original description. However, his description of D. cyanochloris was partly based on specimens of D. humayuni which has a relatively low ventral count (170-178, Vijayakunar & David, 2006). Due to these inaccuracies, the incorrect ventral range of D. formosus (179-205) as given by Boulenger (1894) was nearly identical to the incorrect ventral range of D. cyanochloris (175-207) as given by Wall (1921). In reality, the species differ significantly in ventral count (table 2). In addition, the conspicuous posterior lateral stripes in D. formosus could not be used to differentiate D. formosus from D. cyanochloris as 42 Zootaxa 1394 2007 Magnolia Press VOGEL & ROOIJEN

these were incorrectly thought to be variably present in D. formosus. The fact that such a common snake as D. kopsteini has never been regarded a distinct species must be the result of the longstanding problems surrounding the identities of D. formosus and D. cynochloris: D. kopsteini has simply been hidden from taxonomic herpetology behind a cloud of confusion. The major landmasses of the Sunda region have been interconnected several times during the geological history of the area (Inger & Voris, 2001) enabling species to disperse across the Sunda region. The last connections occurred between 10.000 and 17.000 years ago due to northern glaciation. Especially terrestrial and arboreal species have been able to use these connections to migrate and as a result are frequently wide spread (Inger & Voris, 2001). For example, D. formosus and D. pictus occupy all the major land masses of the Sunda region. D. kopsteini has evidently used those connections to attain its present distribution. However, for some reason, D. kopsteini has not been recorded from Java and Borneo. Java has a relatively small surface area and therefore harbours a relatively low diversity (In Den Bosch, 1985). In addition, its climate is quite different, with a marked dry season throughout much of the island. As such, exclusion of certain species from the Javanese fauna is not surprising. Indeed, two other members of the genus, D. caudolineatus and D. striatus, are absent from the Javanese snake fauna while present in the remainder of the Sunda region which has a wetter or more equatorial climate. More surprising is the absence of D. kopsteini from Borneo. This island has a large surface area and as a result harbours a high biodiversity. Indeed, Borneo is inhabited by all Dendrelaphis species from the Sunda region (D. caudolineatus, D. formosus, D. pictus, D. striatus), with the exception of D. kopsteini. Possibly, this otherwise rather common snake is rare on Borneo, and consequently as yet unrecorded from that island. Specimens examined. Dendrelaphis cyanochloris. North-East India: BMNH 1940.3.4.27, BMNH 1909.3.9.10, BMNH 1940.3.4.26; Burma: BMNH 1946.1.6.13, BMNH 1925.4.2.34; Thailand: PSGV 761b, PSGV 761c small, PSGV 761a large, MNHN 1999.7624, ZMB 58905. Dendrelaphis cf. cyanochloris. Java: RMNH 6880, RMNH 40100, RMNH 7447 (82). Dendrelaphis formosus. Java: SMF 32346, SMF 32347, SMF 32348, MNHN 1975.70, SMF 18667, RMNH 877, RMNH 9005 (1), RMNH 9005 (2), RMNH 40098, RMNH 40105, RMNH 40099, MNHN 7307, MNHN 469, RMNH 40102, RMNH 36443, RMNH 36446, RMNH 40101, RMNH 36444, RMNH 36447, RMNH 36445, ZMB 7151; Peninsular Thailand: BMNH 1940.3.4.28; Peninsular Malaysia: PSGV 793a small, PSGV 782, PSGV 793b large; Borneo: RMNH 8208 (59), RMNH 8208 (61), RMNH 8208 (58), RMNH 8249, RMNH 872; Mentawei-Islands, Sumatra: RMNH 5524; Padang, Sumatra?: RMNH 870; Deli, Sumatra: ZMA 9441. Dendrelaphis humayuni. Nicobars: BMNH 1940.3.4.28. Dendrelaphis kopsteini sp. nov. Peninsular Malaysia: MNHN 1962.1051, ZMH R08445 (former PSGV 269), RMNH 40117 (former PSGV 777), MNHN 1962.1052; Singapore: MNHN 1893.147; Peninsular Thailand: MNHN 1997.6579, QSMI Q 531 (former PSGV 45); Sumatra: ZSM 277/1909 Acknowledgments We are grateful to Patrick David (Paris, France), for his critical reading of this manuscript and his constructive comments that greatly improved the draft of this paper. We thank Annemarie Ohler and Alain Dubois (Paris, France), Wolfgang Böhme (Bonn, Germany), Colin J. McCarthy (London, United Kingdom), Monika Laudahn and Gunther Köhler (Frankfurt am Main, Germany), Frank Glaw and Dieter Fuchs (Munich, Germany), Pim Arntzen and Koos van Egmond (Leiden, The Netherlands), Ronald Vonk and Dik Iliohan (Amsterdam, The Netherlands) and Detlef Langer and Rainer Günther (Berlin, Germany) for their assistance in sending us or letting us examine preserved specimens. Furthermore, we like to thank Ben Horsten (Tilburg, The Netherlands) and Nico van Rooijen (Best, The Netherlands) for translating Boie s original description of D. formosus. Last but not least, we like to thank Myriam A NEW DENDRELAPHIS (SERPENTES: COLUBRIDAE) Zootaxa 1394 2007 Magnolia Press 43