Taxonomic revision of the Neotropical water snake Hydrops triangularis (Serpentes, Colubridae) Author De, Nelson R. Author Lema, Thales De text Zootaxa 2008 1685 55 66 journal article 10.5281/zenodo.180499 a5dfe36d-abca-412b-9346-e014b407023a 1175-5326 180499 Hydrops triangularis Elaps triangularis Wagler, 1824 , in Spix, Sp. Nov. Serp. Bras.: 5, pl. 2a, fig. 1 Type-locality: Ega, lake Tef, at confluence with Rio Solimes, Amazonas, Brazil . Hydrops triangularis ; Wagler, 1830, Nat. Syst. Amph.: 170. Homalopsis Martii (partim); Schlegel, 1837 , Ess. Phys. Serp., 2, p. 356, pl. 13, figs. 19 and 20; t. 1, p. 173. Hydrops martii ; Gray, 1842, Zool. Misc., p. 68. Higina fasciata Gray, 1849, Cat. Sp. Sn., Brit. Mus. p. 75. Hydrops Martii (partim); Dumril, Bibron and Dumril, 1854, Erp. Gn., 7, p. 484. Hydrops Martii ; Lichtenstein , 1856, Nomencl. Rept. Amph. Mus. Zool. Berl., p. 28. Calopsma Martii (partim); Jan, 1865, Arch. Zool. Anat. Phys., 3, p. 242. Hydrops Martii ; Gnther, 1868 , Ann. Mag. Nat. Hist., 4(1), p. 241. Hydrops triangularis ; Boulenger, 1894 , Cat. Sn. Brit. Mus., 2, p. 187, 359. Hydrops triangularis (partim); Boettger, 1898, Kat. Rept. Mus. Senckenb., t. 2, p. 70. Pseuderyx triangularis ; Bocourt, 1895, Miss. Sc. Mex . Rept., p. 806. Pseuderyx inagnitus ; Bocourt, 1899, Le Natur., p. 155. Hydrops triangularis triangularis ; Amaral, 1930 (1929), Mem. Inst. Butantan, 4, p. 92. Hydrops triangularis fasciatus ; Roze, 1957 , Acta Biol. Venez., p. 76. Hydrops triangularis venezuelensis Roze, 1957 , Acta Biol. Venez., p. 78. Hydrops triangularis neglectus Roze, 1957 , Acta Biol. Venez., p. 81. Hydrops triangularis bassleri Roze, 1957 , Acta Biol. Venez., p. 83. Hydrops triangularis bolivianus Roze, 1957 , Acta Biol. Venez., p. 86. Diagnosis . Head not conspicuously distinct from the body. Internasal one. Loreal absent. Dorsal scales smooth, in 15-15-15 rows, lacking apical pits. Anal plate divided. Subcaudals paired. Pupil round. Ventrals 144-185 in males (= 164 ± 7.5, n = 81), 152-183 in females (= 163.6 ± 6.8, n = 88). Total ventrals (venbtrals plus subcaudals) 196-245 in males (= 221.4 ± 10.7, n = 81), 188-240 in females (= 212.7 ± 10.8, n = 88). Subcaudals 32-71 in males (= 57.4 ± 7.2, n = 81), 31-78 in females (= 49 ± 7.3, n = 88). Supralabials 8/8 (n = 168), 9/8 (n = 2), 8/9 (n = 1) or 8/7 (n = 1), with 4-4 (n = 176), 4-5/4-5 (n = 2) or 4/4-5 (n = 2) entering the orbit. Infralabials 8/8 (n = 161), 9/9 (n = 6), 7/7 (n = 3), 8/9 (n=1), or 8/7 (n = 1), with 4/4(n = 161), 4/3 (n = 2), 3/3 (n = 2), or 4/5 (n = 1) contacting the first genials. Preoculars 1/1 (n = 172). Postoculars 2/2 (n = 171) or 1/1 (n = 1). Anterior temporals 1/1 (n = 171) or 2/2 (n = 1). Posterior temporals 1/1 (n = 168), 2/2 (n = 2), 1/2 (n = 1) or 2/1 (n = 1). Body bands 36-70 in males (= 48 ± 7.3, n = 81), 36-75 in females (= 48 ± 7.3, n = 88). Tail bands 10-24 in males (= 14.9 ± 3.0, n = 81), 8-19 in females (= 12.2 ± 2.5, n = 88). Data from hemipenial and skull morphology can be found in Zaher (1999) and Albuquerque (2002) . The largest male and female we recorded were 752 mm and 806 mm total length, respectively. The smallest specimen measured, a female, had a total length of 188mm . The smallest male with rounded deferent ducts was 345mm in snout-vent length and 90mm tail length (MPEG 13826). The smallest female with well-developed eggs was 391mm in snout-vent length (MPEG 9483). We found significant sexual dimorphism in number of subcaudal scales and tail bands in all analyzed groups. Therefore, males and females were analyzed separately for these characters. The number of ventral scales varied significantly among the following groups: the group from north Brazil was significantly different from samples from Peru (H= 60.2878; P <0.0001) and Guyana (H= 49.5422; P <0.0001); the group from Peru was significantly different from samples of Venezuela (H= 69.4481; P <0.0001) and Bolivia (H= 87.2750; P <0.0001); the sample from Guyana was significantly different from sample of Bolivia (H= 76.5294; P <0.0001). The group from north Brazil was significantly different from samples of Peru (H= 54.1590; P <0.0001) and Guyana (H= 48.1972; P <0.0001) in number of body-bands. The ANOVA revealed significant differences in subcaudals among geographic samples of females (F= 5.3977, P = 0.0004) but not for males (F= 1.3746; P = 0.2428) of the combined groups; females from Venezuela were significantly different from females of north Brazil , Guyana , Trinidad , and Peru . Tail-bands did not reveal differences for females (F= 2.2147, P = 0.0600) and males (F= 2.7031, P = 0.0267) of the combined groups. FIGURE 1. Hydrops triangularis (IBSP 65594). Lajeado, Tocantins, Brazil. Courtesy of Otávio A. V. Marques. Arrows indicate irregular projections on anterior and posterior edges of dorsal portion of black bands. Coloration . Most specimens examined were faded, so that the original color pattern was weak or absent. Roze (1957) and Dixon and Soini (1977) described briefly the color pattern of H . triangularis . Color photographs of living specimens of H . triangularis appear in Boos (2001) and Campbell and Lamar (1989 , 2004 ). The most variable aspect of coloration pattern seems to be related to presence and distribution of dorsal red spots. The red spots cover dorsal scale rows II to V and X to XIV; rows VI to IX are always brown or purplebrown ( Fig. 1 ); on the other hand, the red spots may be present only on posterior third of body (see Campbell & Lamar 2004 : Plate 1136). One specimen from Bolivia , MNKR 3698 ( Fig. 2 ), lacks red spots on its dorsum, which is rather yellowish, has a more contrasting color pattern than a specimen in Campbell and Lamar (1989: Fig. 488) . Campbell and Lamar (1989: Fig. 488) depicted a specimen from Venezuela in which the red ventral coloration extends above the first row of dorsal scales. The head is dark brown, with reddish spots scattered over head ( Dixon & Soini 1977 ), a white spot is present on the nasal scales, and a narrow black band covers the posterior portion of parietal scales and the 8th supralabial. Scales on dorsal surface of head are margined with black. According to Boos (2001) , the belly of H . triangularis is pinkish. The preserved specimens examined have black bands on the body and tail and white interspaces varying from one to three scales in width. Boos (2004) published a picture of a melanic specimen collected in Nariva Swamp, Trinidad . It seems that the young specimens have a more contrasting color pattern than adults ( Boos 2001 ). FIGURE 2. Hydrops triangularis (MNKR 3698). Trinidad, Curichi, Bolivia. Courtesy of Lucindo Gonzales. Arrow indicates faint red spots on dorsal scale row II. Boulenger (1894) described H . triangularis as having “black annuli, which may be interrupted and alternate on the middle dorsal and ventral lines”. We are aware that Boulenger’s description was based on 11 specimens collected in Guyana and Surinam (including the syntypes of Higina fasciata = Roze’s Hydrops triangularis fasciatus ), besides of two other specimens whose localities were not mentioned. His description, however, corresponds exactly to what we have observed for specimens collected outside of Guyana and Surinam . Roze (1957) utilized the width of dorsal bands and their conformation to distinguish the subspecies of H. triangularis . According to Roze (1957) specimens from Venezuela have black bands with projections on posterior edge dorsally. We have noted this condition in other specimens outside of Venezuela such as a specimen collected in Tocantins, north Brazil ( Fig. 1 ). The first black band is the broadest. A female collected in Saint-Eugène, French Guyana (MNHN 1996.4566) has the dorsal pattern similar to that of H . caesurus . It has the dorsal pattern of two longitudinal series of dorsal spots that do not reach the vertebral line and two dorsolateral series of spots intercalated between the dorsal series that are an extension of the ventral bands ( Fig. 3 ). The occurrence of H . caesurus in French Guyana is completely outside the previously known range of this species although we have no doubts about the accuracy of this locality. Ecology . These snakes are relatively common in streams found in open forests and savannas (Rivero- Blanco & Dixon 1978 ; Hoogmoed 1982 ; Murphy 1997 ) and they are found in shady, lentic environments. Their activity periods include evening and night, when their prey items are more likely to be encountered ( Albuquerque & Camargo 2004 ). According to Albuquerque and Camargo (2004) , Hydrops triangularis has a dietary preference for fish, especially Synbranchiformes as well as a high fecundity (8- 34 eggs per female). These snakes were located at 450m elevation ( Campbell & Lamar 1989 ). FIGURE 3. Hydrops caesurus (MNHN 1996.4566). Saint-Eugène. French Guyana. Courtesy of Ivan Ineich and Sébastien Soubzmaigne. Note the series of dorsal spots that do not reach the vertebral line. Remarks . Although the samples did differ in numbers of ventrals and body-band counts, the differences were small ( Fig. 4–5 ; Table 1 ). The ranges of North Brazil group encompass most observed in other groups. Specimens from North Brazil apparently have fewer ventrals than snakes from Peru , and fewer body bands than specimens from western South America ( Venezuela , Ecuador , Colombia , Peru and Bolivia ) ( Table 1 ). Considerable differences exist among the ecological conditions within the distribution of H. triangularis (e.g. Roze 1966 ; Cunha & Nascimento 1978 ; Hoogmoed 1979 , 1982 ; Lancini 1982 ; Murphy 1997 ; Boos 2001 ). According to Roze (1966) , H. t. venezuelensis is distributed in the Gransabánica and Horinocense formations, southern sub-region, which includes all of southern Venezuela . For Roze (1966) , the snake fauna of this sub-region is most closely related to the snakes of Brazilian Amazon and Western Guyana . Hoogmoed (1979) noted that there is a distinct barrier within Guyana formed by the Essequibo Rio Branco Depression, however, he also noted this barrier serve as dispersal route for aquatic species. Based on the statements of Roze (1966) and Hoogmoed (1979) , we infer that the absence of an ecological barrier among populations of Venezuela , Trinidad , Guyana and North Brazil , appear to permit a free gene flow across those populations. Boos (1984) suggested a possible transmigration of reptile specimens from Paria Peninsula, western Venezuela to Trinidad . As most of the reptiles of Trinidad have a South American origin ( Boos 1984 , 2001 ; Hedges 1996 ), possibly H. triangularis dispersed from Venezuela to Trinidad through the Orinoco river Delta, which is situated along coastal Venezuela , south of Trinidad and north of the Guayana highlands. TABLE 1. Mean, standard deviation, and range for all groups of H. triangularis examined in this study. Male: m; Female: f.

North Brazil (n=78)	Trinidad and Brit- Venezuela, Ecu., ish Guyana1 Colombia (n=12) (n=13)	Bolivia (n=12)	Peru (n=20)	E of British Guyana, Suriname, and Fr Guyana (n=34)
Ventrals	m 162.1 ± 6.30 152–181 f 162 ± 6.16 152–183	m 161.3 ± 2.51 m 160.6 ± 5.96 159–164 156–171 f 161.2 ± 5.47 f 160.4 ± 5.57 153–169 153–167	m 156.7 ± 12.5 156–171 f 156.5 ± 3.00 144–175	m 169.3 ± 6.37 162–185 f 175 ± 7.69 167–193	m 167.3 ± 5.42 157–177 f 170.1 ± 2.43 165–175
Subcaudals	m 56.1 ± 6.96 38–67 f 49 ± 6.88 39–78	m 54.3 ± 6.80 m 57 ± 6.03 49–62 42–71 f 51.6 ± 3.77 f 38.8 ± 9.95 48–59 31–47	m 57.1 ± 5.11 42–71 f 41.5 ± 9.99 32–51	m 61.3 ± 6.57 48–70 f 53.1 ± 5.08 44–58	m 58.6 ± 4.91 46–66 f 50.8 ± 6.64 42–67
Body-bands	m 44.5 ± 4.88 36–60 f 44 ± 4.60 36–62	m 52 ± 5.0 m 46.5 ± 12.1 47–57 39–56 f 50 ± 5.52 f 56.8 ± 7.06 43–59 43–75	m 50 ± 11.7 47–62 f 44.2 ± 5.00 41–75	m 52.7 ± 8.18 40–70 f 50 ± 5.00 46–62	m 49.5 ± 6.29 38–60 f 50 ± 4.87 44–58
Tail-bands	m 13.9 ± 2.69 10–19 f 11.5 ± 2.25 8–18	m 13.3 ± 2.30 m 14.6 ± 3.53 12–16 10–20 f 12.5 ± 1.51 f 13.1 ± 4.08 11–15 8–19	m 16.4 ± 3.95 10–20 f 11 ± 2.60 12–24	m 16.3 ± 2.98 11–20 f 12.6 ± 2.60 10–15	m 15.4 ± 2.21 11–20 f 13.5 ± 2.24 11–18

1western of British Guyana ; includes the type-series of H. t. neglectus . The analysis of the characters employed by Roze (1957) for the diagnosis of the subspecies of H. triangularis revealed overlapping values in the number of ventral and subcaudal scales and the number of bands on the body and tail. One of the main characters proposed by Roze (1957) to recognize the subspecies, width of the bands and their conformation on the dorsum was not found to be of taxonomic value in the present study. The differences observed among the black bands of the subspecies of H. triangularis as presented by Roze (1957) are here regarded as falling within the variation of a widely distributed species ( Fig. 6 ). According to Mayr and Ashlock (1991) , most subspecies are geographic isolates and the borders of their ranges are easily established. However, the borders of the subspecies under study were not established because of the total absence of intergrades among the samples. The examined groups revealed only a few quantitative differences among the characters under study. We conclude that H. triangularis should not be divided into subspecies. The subspecies of H. triangularis were poorly distinguished based on the number of ventral and subcaudal scales as well as the number of bands on the body and tail. We suggest that these characters were not useful to define the subspecies of H . triangularis .