Systematics of South American snail-eating snakes (Serpentes, Dipsadini), with the description of five new species from Ecuador and Peru

Abstract A molecular phylogeny of the Neotropical snail-eating snakes (tribe Dipsadini) is presented including 43 (24 for the first time) of the 77 species, sampled for both nuclear and mitochondrial genes. Morphological and phylogenetic support was found for four new species of Dipsas and one of Sibon, which are described here based on their unique combination of molecular, meristic, and color pattern characteristics. Sibynomorphus is designated as a junior subjective synonym of Dipsas. Dipsas latifrontalis and D. palmeri are resurrected from the synonymy of D. peruana. Dipsas latifasciata is transferred from the synonymy of D. peruana to the synonymy of D. palmeri. A new name, D. jamespetersi, is erected for the taxon currently known as Sibynomorphus petersi. Re-descriptions of D. latifrontalis and D. peruana are presented, as well as the first photographic voucher of an adult specimen of D. latifrontalis, along with photographs of all known Ecuadorian Dipsadini species. The first country record of D. variegata in Ecuador is provided and D. oligozonata removed from the list of Peruvian herpetofauna. With these changes, the number of Dipsadini reported in Ecuador increases to 22, 18 species of Dipsas and four of Sibon.


Introduction
With 70 currently recognized species (Table 1), the snail-eaters (tribe Dipsadini) are among the most diverse groups of arboreal snakes (Wallach et al. 2014;Uetz et al. 2016). Some authors have suggested that their tree-dwelling lifestyle and specialized diet resulted this large an adaptive radiation (e.g., MacCulloch and Lathrop 2004;Sheehy 2012). In the last decade, the limits of the tribe have been redefined to include five genera (Dipsas, Plesiodipsas, Sibon, Sibynomorphus, and Tropidodipsas;), but recent studies suggest that not all of them are monophyletic (Sheehy 2012;Figueroa et al. 2016). Consequently, the limits between genera, species, and species groups appear to be poorly defined, and in need of revision for a robust and stable taxonomy.
One of the first modern attempts to clarify the taxonomy and summarize knowledge on the tribe Dipsadini was published by Peters (1960). Peters considered Dipsadini to include the genera Dipsas, Sibon and Sibynomorphus. Later, Zaher (1999) and  added Tropidodipsas and Plesiodipsas in the tribe. Peters also created seven species groups within Dipsas, three within Sibon (Table 1), and recognized D. boettgeri, D. latifrontalis, D. latifasciata, D. polylepis, and D. peruana as distinct species based on coloration and lepidosis. However, he considered D. palmeri and D. praeornata to be synonyms of D. latifrontalis.
After Peters, several authors continued to address the systematics of the group (Downs 1961, Hoge 1964, Peters and Orejas-Miranda 1970, Kofron 1982, Orcés and Almendáriz 1987, Porto and Fernandes 1996, Fernandes et al. 1998, Fernandes et al. 2002, Cadle and Myers 2003, Passos et al. 2004, Passos et al. 2005, Cadle 2005, Cadle 2007, Harvey 2008, Harvey and Embert 2008. Of these, the works by Cadle and Myers (2003), Cadle (2007), Harvey (2008), and Harvey and Table 1. Taxonomy of Dipsadini prior to this paper. Embert (2008) are worth addressing further because they focused on Ecuadorian species for which there is still taxonomic uncertainty. Cadle and Myers (2003) removed D. variegata from the herpetofauna of Ecuador, since previous records were based on museum misidentifications. Cadle (2007) reviewed the status of species of Sibynomorphus in Ecuador and Peru, and referred three additional specimens (AMNH 110587, BMNH 1935.11.3.108, andMUSM 2192) to S. oligozonatus, including the first country record for Peru. Cadle (2005) also reviewed three specimens of D. gracilis collected in Peru; however, Harvey (2008) concluded that only one of them corresponded to D. gracilis. In the same work, Harvey also redefined Peters' (1960) species groups (Table 1). Finally, Harvey and Embert (2008) transferred D. boettgeri, D. latifrontalis, and D. polylepis to the synonymy of D. peruana, based on both the difficulty of segregating these species using morphological characters and their "more or less continuous distribution along the eastern slopes of the Andes".
Here, we combine morphological analysis and molecular phylogenetics to revise generic and species limits within Dipsadini. We combine all available molecular sampling with new samples from Ecuador, Peru, Brazil and Costa Rica, and find support for five new species, as well as a number of changes to the geographic distribution of several Andean species.

Ethics statement
This study was carried out in strict accordance with the guidelines for use of live amphibians and reptiles in field research (Beaupre et al. 2004) compiled by the American Society of Ichthyologists and Herpetologists (ASIH), the Herpetologists' League (HL) and the Society for the Study of Amphibians and Reptiles (SSAR). All procedures with animals (see below) were reviewed by the Ministerio de Ambiente del Ecuador (MAE) and specifically approved as part of obtaining the following field permits for research and collection:

Common names
Criteria for common name designation are as proposed by Caramaschi et al. (2006), as modified by Coloma andGuayasamin (2011-2017), and are as follows (in order of importance): (i) the etymological intention (implicit or explicit) that the authors used when naming the species (specific epithet); (ii) a common name that is already widely used in the scientific literature; (iii) a common name that has an important ancestral or cultural meaning; (iv) a common name based on any distinctive aspect of the species (distribution, morphology, behavior, etc.).

Laboratory techniques
Genomic DNA was extracted from 96% ethanol-preserved tissue samples (liver, muscle tissue or scales) using either a guanidinium isothiocyanate extraction protocol, or a modified salt precipitation method based on the Puregene DNA purification kit (Gentra Systems). We amplified the 16S gene using primer pairs 16Sar-L / 16Sbr-H-R from Palumbi et al. (1991) and 16sF.0 (Pellegrino et al. 2001) / 16sR.0 (Whiting et al. 2003). Additionally, the Cytb gene was obtained with primer pairs GLUDG-L (Palumbi et al. 1991) / ATRCB3 (Harvey et al. 2000) and LGL765 (Bickham et al. 1995) / CytbV (Torres-Carvajal et al. 2015), whereas the gene coding for the subunit 4 of the NADH dehydrogenase was amplified with the primers ND4 and Leu developed by Arévalo et al. (1994). The c-mos gene was retrieved with primers S77 and S78 developed by Lawson et al. (2005). PCR reactions contained 2 mM (Cytb and ND4) or 3 mM (16S and c-mos) MgCl 2 , 200 µM dNTP mix, 0.2 µM (16S, Cytb and c-mos) or 0.8 µM (ND4) of each primer and 1.25 U (16S) or 0.625 U (ND4, Cytb and cmos) Taq DNA Polymerase Recombinant (Thermo Fisher Scientific) in a 25 µL total volume. The nucleotide sequences of the primers and the PCR conditions applied to each primer pair are detailed in Appendix 2. PCR products were cleaned with either ExoSAP-IT (Affymetrix, Cleveland, OH), or Exonuclease I and Alkaline Phosphatase (Illustra ExoProStar by GE Healthcare) before they were sent to Macrogen Inc (Korea) for sequencing. All PCR products were sequenced in both forward and reverse directions with the same primers that were used for amplification. The edited sequences were deposited in GenBank (Appendix 1).

DNA sequence analyses
A total of 298 DNA sequences were used to build a phylogenetic tree of the tribe Dipsadini, of which 222 were generated during this work and 76 were downloaded from GenBank. Among the new sequences, 103 are 201-520 bp long fragments of the 16S gene, 91 are 586-1,090 bp long fragments of the Cytb gene, 45 are 443-583 bp long fragments of the c-mos gene, 31 are 242-473 bp long fragments of the 12S gene, and 28 are 593-699 bp long fragments of the ND4 gene. New sequences were edited and assembled using the program Geneious ProTM 5.4.7 (Drummond et al. 2010) and aligned with those downloaded from GenBank (Appendix 1) using MAFFT v.7 (Katoh and Standley 2013) under the default parameters in Geneious ProTM 5.4.7. Genes were combined into a single matrix with 11 partitions, one per non-coding gene and three per protein-coding gene corresponding to each codon position. The best partition strategies along with the best-fit models of evolution were obtained in PartitionFinder 2 (Lanfear et al. 2016) under the Bayesian information criterion.
Phylogenetic relationships were assessed under both a Bayesian inference (BI) and a maximum likelihood (ML) approach in MrBayes 3.2.0 (Ronquist and Huelsenbeck 2013) and RAxML v8.2.9 (Stamatakis 2006), respectively. For the ML analysis, nodal support was assessed using the rapid-bootstrapping algorithm with 1000 non-parametric bootstraps. All ML estimates and tests were run under the GTRCAT model, as models available for use in RAxML are limited to variations of the general time-reversible (GTR) model of nucleotide substitution. For the BI analysis, four independent analyses were performed to reduce the chance of converging on a local optimum. Each analysis consisted of 6,666,667 generations and four Markov chains with default heating settings. Trees were sampled every 1,000 generations and 25% of them were arbitrarily discarded as ''burn-in." The resulting 5,000 saved trees per run were used to calculate posterior probabilities (PP) for each bipartition in a 50% majority-rule consensus tree. We used Tracer 1.6 (Rambaut et al. 2018) to assess convergence and effective sample sizes (ESS) for all parameters. Additionally, we verified that the average standard deviation of split frequencies between chains and the potential scale reduction factor (PSRF) of all the estimated parameters approached values of ≤0.01 and 1, respectively. Genetic distances between new species and their closest morphological relative were calculated using the uncorrected distance matrix in PAUP 4.0 (Swofford 2002). GenBank accession numbers are listed in Appendix 1.

Morphological data
Terminology for Dipsadini cephalic shields follows proposals by Peters (1960) and Harvey and Embert (2008). Diagnoses and descriptions generally follow Fernandes et al. (2010), and ventral and subcaudal counts follow Dowling (1951). When providing the standard deviation, we use the ± symbol. We examined comparative alcoholpreserved specimens from the herpetology collections at Museo de Zoología de la Universidad Tecnológica Indoamérica (MZUTI), Museum d'Histoire Naturelle de la Ville de Genève (MHNG), Museo de Zoología de la Pontificia Universidad Católi-ca del Ecuador (QCAZ), National Museum of Natural History (USNM), División de Herpetología del Instituto Nacional de Biodiversidad (DHMECN), Museo de Zoología de la Universidad del Azuay (MZUA), American Museum of Natural History (AMNH), Museo de Zoología de la Universidad San Francisco de Quito (ZSFQ), Museum of Natural Science of the Louisiana State University (LSUMZ), Museum of Comparative Zoology of Harvard University (MCZ), Natural History Museum and Biodiversity Research Center of University of Kansas (KU), British Museum of Natural History (BMNH), Museo de Historia Natural de la Escuela Politécnica Nacional (EPN), and Museo de la Universidad Nacional de San Marcos (MUSM) ( Table 2). Morphological measurements were taken with measuring tapes to the nearest 1 mm, or with digital calipers to the nearest 0.1 mm. Abbreviations are as follows: snoutvent length (SVL); tail length (TL). Sex was determined by establishing the presence/ absence of hemipenes through a subcaudal incision at the base of the tail unless hemipenes were everted.

Molecular phylogeny and taxonomic consequences
We consider strong support to be bootstrap values of >70% and posterior probability values >95% following Felsenstein (2004). Overall, there is low support for the relationship between the genera Dipsas, Sibon, and Tropidodipsas (Fig. 3). The genus Sibynomorphus is not monophyletic and the included species are nested in four mutually exclusive clades within Dipsas. Two of the three included species of Tropidodipsas, T. fischeri, and T. fasciata, form a poorly supported clade, whereas T. sartorii is strongly supported as sister lineage to all other included samples of Dipsadini. The genus Sibon is monophyletic, and sister to T. fischeri and T. fasciata in the ML analysis, although this relationship is not strongly supported. In the BI analysis, Sibon is sister to Dipsas. We excluded Sibon noalamina (voucher SMF 91539) from the analyses as the short sequence available in GenBank (gene fragment 16S) represented a rogue taxon that assumed varying phylogenetic positions in the tree collection used to build the consensus tree.
Sibon longifrenis is recovered as the sister taxon to all other included species of Sibon. Deep intraspecific divergence is found between samples of S. annulatus from Central America (MVZ 269290, ADM 0007, ADM 242) and that from Ecuador (MZUTI 3034). The widespread species S. nebulatus is paraphyletic with respect to both S. dunni and a new species from Ecuador. Nonetheless, within S. nebulatus, the included subspecies S. n. nebulatus (Linnaeus, 1758) and S. n. leucomelas (Boulenger, 1896) are monophyletic, while the single Colombian specimen of S. n. hartwegi (Peters 1960) is sister to all other members of the Ecuadorian S. nebulatus group. However, posterior probabilites from our genetic data for the formation of monophyletic Ecuadorian clades S. n. leucomelas, S. dunni, and Sibon. sp. are variable, and as low as 48% PP for the node separating Sibon sp. from S. nebulatus leucomelas and S. dunni. Table 2. Locality data for specimens examined in this study. Coordinates represent actual GPS readings taken at the locality of collection or georeferencing attempts from gazetteers under standard guidelines, though some variation from the exact collecting locality will be present. Similarly, elevations are taken from Google Earth, and may not exactly match the elevations as originally reported. Specimens listed here but not under Appendix 3 were examined indirectly (e.g., through photographs). Eight Sibynomorphus species were included in the molecular analyses. These are S. mikanii, S. neuwiedi, S. oligozonatus, S. petersi, S. turgidus, S. vagus, S. ventrimaculatus, and S. williamsi. In the ML analysis, all of them are nested within different Dipsas subclades, whereas in the BI analysis, the clade containing S. mikanii and S. turgidus is not nested within Dipsas. Crucially, Dipsas mikanii Schlegel, 1837 is the type species of Sibynomorphus (Fitzinger, 1843). Thus, we synonymize Sibynomorphus with Dipsas primarily based on the ML analysis, which mirrors the results of Sheehy (2012).

Species
Based on our transfer of the genus Sibynomorphus Fitzinger to the synonymy of Dipsas, we propose the following binomial nomenclature for the eleven species traditionally included in the genus Sibynomorphus: Dipsas lavillai comb. n., D. mikanii, D. neuwiedi comb. n., D. oligozonata comb. nov., D. oneilli comb. n., D. turgida comb. nov., D. vagrans comb. n., D. vaga comb. n., D. ventrimaculata comb. n., and D. williamsi comb. n. However, we refrain from applying D. "petersi" for Sibynomorphus petersi here, because the name Dipsas "indica" petersi (Hoge & Romano, 1975), another taxon and putative species from southeastern Brazil, is often already named as Dipsas petersi (e.g., Centeno et al. 2008, Wallach et al. 2014, and this name predates Sibynomorphus petersi (Orces & Almendáriz, 1989). Therefore, the latter is now a secondary junior homonym in conflict upon transfer to Dipsas Laurenti, and thus requires a replacement name. We therefore erect the name Dipsas jamespetersi, which still honors James A. Peters, for the taxon Sibynomorphus petersi Orces & Almendariz, 1989. There are several clades within Dipsas peruana sensu lato. One is D. peruana, the other is a new species from northern Ecuador, which we describe below, and the third is the lineage corresponding to the population distributed along the Amazonian slopes of the Andes between central Ecuador and northern Peru. Below, we resurrect the name D. palmeri (Boulenger, 1912) for this lineage, as the type locality of D. palmeri (El Topo, province of Tungurahua, Ecuador) is located within the geographic range of the included samples (Fig. 4) and the holotype agrees in coloration and lepidosis with other specimens (Appendix 3) in the same region that were included in the genetic analyses.
Dipsas oligozonata is the strongly supported sister lineage of a clade that includes three species: D. williamsi and two new species from western Ecuador and northern Peru, which we describe below. Dipsas indica is paraphyletic with respect to D. bucephala. Dipsas jamespetersi is paraphyletic with respect to a sample of D. vaga (KU 219121).
Based on the species included in the phylogenetic analysis, the Dipsas articulata and D. indica groups, sensu Harvey 2008 (Table 1), are recovered as monophyletic. The other groups included in the phylogenetic analysis (i.e., catesbyi, oreas, pratti, temporalis and variegata) are not monophyletic. The two included members of the D. catesbyi group (i.e., D. catesbyi and D. pavonina) are not sister taxa. The included members of the Dipsas oreas group form a paraphyletic unit, because besides including D. elegans, D. ellipsifera, and D. oreas, this group also includes D. andiana, a species that was considered a member of the D. variegata group   Table 1). Accordingly, we transfer D. andiana to the D. oreas group. The two included members of the D. pratti group (i.e., D. peruana and D. pratti) are placed in different branches of the phylogeny. The same is true for the included members of the D. temporalis group (i.e., D temporalis and D. vermiculata), whereby D. vermiculata clusters with D. variegata, and accordingly we move it into that group. We refrain from merging the Dipsas temporalis and D. pratti groups because we did not examine the specimens of D. pratti included in the analysis (MHUA 14278). We also refrain from assigning further species groups until a more complete taxon sampling is made available.

New records for Ecuador
One individual (Fig. 1v) of Dipsas variegata photographed (not collected) at Gareno Lodge, province of Napo (S1.03559, W77.39864; 336 m), represents the first record of this species in Ecuador (Fig. 4). This individual agrees in coloration with the description of the species provided by Cadle and Myers (2003) and Mebert et al. (submitted), including dorso-lateral blotches/saddles resembling vertically stretched rhomboids or bars, often with a light center or spots, border of blotches being zig-zag shaped and following the outline of adjacent dorsal scales, variably numbered and shaped spots in the interspaces, cephalic blotches lacking yellow borders, and a light-colored eye. It shows also the typical truncated head (see Peters 1960 for description of head truncation) of D. variegata, in particular the short, but high preorbital region including an upturned chin, a convex supraocular, narrow and vertically elongated anterior labials (here 2 nd -6 th supralabials), and 15 dorsal scale rows. This D. variegata expands the known distribution 1,186 km SW from the nearest localities along the Venezuelan Andes (Natera-Mumaw et al. 2015) and 1,343 km NW from the nearest locality in southeastern Peru (Catenazzi et al. 2013).

Systematic accounts
We seek here to name or provide re-descriptions only for species that are monophyletic in our molecular phylogeny and share diagnostic features of their coloration pattern and lepidosis. Based on these species delimitation criteria, which follow the general species concept of de Queiroz (2007)    Diagnosis. Sibon bevridgelyi is placed in the genus Sibon based on phylogenetic evidence (Fig. 3) and on having the labial beneath primary temporal conspicuosly higher than other labials. The species differs from all described species of Sibon based on the following combination of characters: (1) 15/15/15 smooth dorsals with enlarged vertebral row (1.3-1.7 times as wide as adjacent rows); (2) seven supralabials with 4 th and 5 th contacting orbit or eight supralabials with 5 th and 6 th contacting orbit; (3) one pair of infralabials in contact behind symphysial; (4) postmental absent; (5) 175-193 ventrals in males, 193 in the single female; (6) 80-94 divided subcaudals in males, 98 in the single female; (7) dorsal and ventral ground color pale yellow with or without irregular black bands, and with a black stippled disruptive pattern of irregular rusty to reddish brown blotches that are separated from each other by light interspaces (Figs 6, 2b, c); bands incomplete and stippling not prominent or absent on ventral surfaces; head heavily speckled or blotched with black or rusty pigment; eyes light slate blue to pale goldenrod with black speckles and reticulations; (8) 349-732 mm SVL in males, 786 mm in the single female; (9) 124-268 mm TL in males, 204 mm in the single female.
Comparisons. Sibon bevridgelyi is most similar to S. nebulatus, from which it differs on the basis of its distinctive coloration (Figs 6, 2b, c). In S. nebulatus (Figs 2e, f ), the dorsal and ventral color is a combination of mainly black to dark-brown blotches or bands on a gray to grayish brown background (interblotch) color; the dorso-lateral blotches can partly be bordered by white to rosy scales or edges. In some regions, the blackish pattern and gray ground color is often replaced by dark and light brown tones (e.g., in Venezuela, adjacent regions in Colombia, and Trinidad and Tobago); the spaces between the blotches are heavily invaded by blotch color and strongly stippled, spotted and mottled with white and black pigment. Although S. bevridgelyi also has a disruptive pattern, the diagnostic white and gray pigment of S. nebulatus from Central America and northern South America is lacking in S. bevridgelyi. Instead of white pigment, there is golden yellow; instead of gray, the dominant ground color is bright rusty brown to maroon. Additionally, the infralabials and the whitish throat in S. nebulatus from Central America and northern South America are heavily stippled or at least partly interrupted with black pigment, whereas in S. bevridgelyi the infralabials and the throat are immaculate or have few scattered blotches (Fig. 7b). Finally, the black blotches and stippling diagnostic of S. nebulatus are lacking in the majority of the specimens of S. bevridgelyi. Specimens of S. nebulatus with rosy gray or reddish brown ground color have white (instead of yellowish) blotches and stippling. Genetic divergence in a 521 bp long fragment of the mitochondrial Cytb gene between S. bevridgelyi and S. nebulatus leucomelas is 1.9-2.5%, whereas intraspecific distances are less than 0.4% in both species.
Description of holotype. Adult male, SVL 602 mm, tail length 186 mm (31% SVL); head length 20.9 mm (3% SVL) from tip of snout to commissure of mouth; head width 12.4 mm (59% head length) taken at broadest point; snout-orbit distance 21 mm; head distinct from neck; snout short, blunt in dorsal and lateral outline; rostral 3.5 mm wide, broader than high; internasals 1.9 mm wide, broader than long; prefrontals 4.4 mm wide, longer than broad, entering orbit; supraocular 4.4 mm long, longer than broad; frontal 4.1 mm long, pentagonal and rounded, in contact with prefrontals, supraoculars, and parietals; parietals 7.7 mm long, longer than broad; nasal weakly divided, in contact with first three supralabials, loreal, prefrontal, internasal, and rostral; loreal 3.7 mm long, longer than high, entering the orbit; eye diameter 3.9 mm; pupil semi-elliptical; no preocular; two postoculars; temporals 1+3 on the right side, 2+3 on the left side; eight supralabials with 5 th and 6 th contacting orbit on the right side, seven supralabials with 4 th and 5 th contacting orbit on the left side; symphysial separated from chinshields by the first pair of infralabials; nine infralabials, 1-7 contacting chinshields; anterior pair of chinshields broader than long, posterior pair longer than broad; dorsal scales in 15/15/15 rows, smooth, without apical pits; 184 ventrals; 80 divided subcaudals; cloacal plate single.
Natural history. Specimens of Sibon bevridgelyi have been found active at night (20h56-03h56) on arboreal vegetation 30-500 cm above the ground in secondary and primary semideciduous foothill forest, pastures, and cacao plantations, usually close to streams. QCAZ 14444 was found feeding on a snail. In captivity, MZUA.RE.0142 fed on slugs and snails. By daytime, one individual (not collected) was found hidden under tree bark, and another (ZSFQ D503) was found coiled on the center of a palm tree about 2 m above the ground. DHMECN 9483 was collected in sympatry with Dipsas andiana and D. bobridgelyi at Reserva Biológica Buenaventura.
Distribution. Northwestern Peru in the department of Piura, and southwestern Ecuador in the provinces of Azuay, Chimborazo, El Oro, Guayas, Los Ríos and Mana-bí at elevations between 5 and 1206 m (Fig. 8).
Etymology. The specific epithet honors the late Prof. Beverly S. Ridgely, life-long birder and conservationist, and father of Robert S. Ridgely, well known in Ecuadorian ornithological circles and co-author of The Birds of Ecuador. Though he never got to visit Buenaventura, from afar Bev continued to delight in the conservation successes of Fundación Jocotoco, which now owns and manages one of the few protected areas where the Vulnerable Sibon bevridgelyi is known to occur.
Conservation status. We consider Sibon bevridgelyi to be Vulnerable following B2a,b(i,iii) IUCN criteria (IUCN 2001) because its area of occupancy is estimated to be less than 2,000 km 2 , it is known only from 15 patches of forest lacking connectivity between them, and its habitat is severely fragmented and declining in extent and quality due to deforestation. Furthermore, only three of the localities (Parque Nacional Machalilla, Reserva Buenaventura, and Reserva Ayampe) where S. bevridgelyi occurs are currently protected. Diagnosis. Dipsas bobridgelyi is placed in the genus Dipsas based on phylogenetic evidence (Fig. 3), and the absence of a labial that is noticeably higher than other labials and in contact with the postocular, primary, and secondary temporals. The species differs from all described species of Dipsas based on the following combination of characters: (1) 15/15/15 smooth dorsals with enlarged vertebral row (2.1-2.2 times as wide as adjacent rows); (2) loreal and prefrontal in contact with orbit; (3) 9 supralabials with 4 th and 5 th contacting orbit; (4) one pair of infralabials in contact behind symphysial; (5) 180-201 ventrals in males, 178-184 in females; (6) 95-117 divided subcaudals in males, 96-98 in females; (7) dorsal and ventral color made up of 30-35 bold black body rings (up to 7-12 vertebral scales long) separated from each other by narrow (up to 3-4 vertebral scales long) dingy white interspaces; dorsal aspect of interspaces heavily speckled with rusty and black pigment; ventral surfaces lacking speckling; ground color of head dingy white with various degrees of scattered black pigment that coalesce on the top of the head, and various degrees of rusty speckling concentrated on the snout, nape and sides of the head; iris rich dark brown; (8) 372-478 mm SVL in males, 286-404 mm in females; (9) 158-212 mm TL in males, 117-158 mm in females.
Comparisons. Dipsas bobridgelyi is most similar to D. gracilis, from which it differs in coloration. In D. gracilis (Figs 1h, i), the black rings are up to 10-16 vertebral scales long and the interspaces are up to 5-7 scales long, whereas in D. bobridgelyi the black rings and interspaces are shorter, up to 8-9 and 3-4 vertebral scales long, respectively. In D. gracilis, the head plates are either completely black or black scattered with reddish brown, whereas in D. bobridgelyi the head plates are heavily stippled with white and tan pigment, especially on the prefrontals and internasals. In all known specimens of D. bobridgelyi, the ground color of the interspaces is white with contrasting reddishtan pigment in the center, whereas in D. gracilis the ground color of the light interspac-es on body and tail is either completely light brown or light reddish white, gradually becoming reddish brown towards the center. Finally, the nape and temporal region of the head in D. gracilis are either immaculate light reddish brown or marked with bold black speckles, whereas in D. bobridgelyi they are an irregular mix of fine speckling of white, rusty, and black pigments. Genetic divergence in a 689 bp long fragment of the mitochondrial Cytb gene between D. bobridgelyi and D. gracilis is 8.7-9.0%, whereas intraspecific distances are less than 0.3% in both species.
Natural history. Individuals of Dipsas bobridgelyi have been found active at night (19h00-23h26) on arboreal vegetation 100-250 cm above the ground in secondary semi-deciduous foothill forest. MZUTI 5414 was found feeding on a snail.
Distribution. Foothills of the southwestern Ecuadorian Andes in the provinces of Azuay and El Oro, and northwestern Peruvian Andes in the department of Tumbes, at elevations between 39 and 572 m (Fig. 4).
Etymology. This species is named in honor of Dr. Robert "Bob" S. Ridgely, a leading ornithologist and distinguished conservationist who has dedicated almost 50 years of his life to the study and conservation of birds and biodiversity across Latin America. Bob is the President of Rainforest Trust and for the past twenty years has been a major driver of conservation in Ecuador through Fundación Jocotoco, which he helped establish twenty years ago. In 1980, Bob visited the type locality of Dipsas bobridgelyi (Buenaventura, meaning "good fortune"), now known to be a key area for the conservation of biodiversity. Bob embarked on conservation and worked diligently to raise funds through Rainforest Trust for the past 18 years to purchase private properties and establish what is now the Reserva Buenaventura of Fundación Jocotoco.
Conservation status. We consider Dipsas bobridgelyi to be Endangered following the IUCN criteria B1a,b(i,iii) (IUCN 2001) because its extent of occurrence is estimated to be less than 5,000 km 2 , it is known only from 4 patches of forest lacking connectivity between them, and its habitat is severely fragmented and declining in extent and quality due to deforestation. Furthermore, only two of the localities (Buenaventura reserve and Reserva Nacional de Tumbes) where D. bobridgelyi occurs are currently protected.
Remarks. Cadle (2005) and Harvey (2008) examined MUSM 17589 from Tumbes department, Peru, and concluded that it was Dipsas gracilis. Although we did not examine this specimen, we believe that it corresponds to D. bobridgelyi based on Cadle's (2005) color description (i.e., head white with many irregular black markings on the top and sides). Diagnosis. Dipsas georgejetti is placed in the genus Dipsas based on phylogenetic evidence (Fig. 3) and the absence of a labial that is noticeably higher than other labials and in contact with the postocular, primary and secondary temporals. The species differs from all described species of Dipsas based on the following combination of characters: (1) 15/15/15 smooth dorsals with a slightly enlarged vertebral row (1-1.4 times as wide as adjacent rows); (2) loreal and prefrontal in contact with orbit; (3) 7 supralabials with 4 th and 5 th (3 th -5 th in DHMECN 11646) contacting orbit; (4) no infralabials in contact behind symphysial; (5) 172-180 ventrals in males, 177 in one female; (6) 69-86 divided subcaudals in males, 58 in one female; (7) dorsal ground color light sandy brown with a pattern of 53-61 drab to brown black-edged middorsal blotches that are wider (6-7 vertebral scales long) and solid down to the edges of the ventrals on the first one third of the body, but becoming narrower (1-3 vertebral scales long) and broken up laterally towards the tail; interspaces finely speckled with brown pigment; ground color of the head light sandy brown with bold dark brown to black irregular blotches scattered on head plates and edging supralabials; ventral surfaces sandy brown with fine black speckling; iris sandy brown with dense dark brown speckling; (8) 270-711 mm SVL in males, 856 mm in one female; (9) 87-170 mm TL in males, 150 mm in one female.

Dipsas georgejetti
Comparisons. Dipsas georgejetti is most similar to D. oswaldobaezi, D. williamsi, D. oligozonata, and D. vagrans, in that order, all of which were previously included in the genus Sibynomorphus. From D. oswaldobaezi (Figs 13, 14) and D. williamsi, it differs in having 7 supralabials with 4 th and 5 th bordering the eye (instead of 6 with 3 rd and 4 th bordering the eye). It further differs from D. williamsi in having the first supralabial not in contact with prefrontal (vs. in broad contact in D. williamsi). From D. oligozonata (Fig. 1o) and D. vagrans, it differs in having more than 160 ventrals. Dipsas georgejetti further differs from D. oligozonata in having distinct bold crossbands at least middorsally along the whole length of the body, instead of being present only on the anterior half of the body. Genetic divergence in a 529 bp long fragment of the mitochondrial Cytb gene between D. georgejetti and D. oswaldobaezi is 8.3%, whereas intraspecific distances are less than 0.4% in D. georgejetti. For the same fragment, the distance between D. georgejetti and D. williamsi is 7.8-7.9%.
Description of holotype. Adult male, SVL 315 mm, TL 87 mm (28% SVL); head length 13.6 mm (4% SVL) from tip of snout to commissure of mouth; head width 8.4 mm (62% head length) taken at broadest point; snout-orbit distance 3.5 mm; head distinct from neck; snout short, blunt in dorsal and lateral outline; rostral 2.0 mm wide, broader than high; internasals 1.7 mm wide, broader than long; prefrontals 2.5 mm wide, longer than broad and contacting orbit; supraocular 3.4 mm long, longer than broad; frontal 3.3 mm long, pentagonal, in contact with prefrontals, supraoculars, and parietals; parietals 5.5 mm long, longer than broad; nasal divided, in contact with first two supralabials, loreal, prefrontal, internasal, and rostral; loreal 1.7 mm long, slightly higher than long, entering orbit; eye diameter 2.8 mm; pupil semielliptical; no preocular; two postoculars; temporals 2+2; seven supralabials, 4 th and 5 th contacting orbit; symphysial in contact with first pair of chinshields; nine infralabials, 1-6 contacting chinshields; anterior pair of chinshields longer than broad, posterior pair broader than long; dorsal scales in 15/15/15 rows, smooth, without apical pits; 178 ventrals; 69 divided subcaudals; cloacal plate single. Natural history. The holotype was active during a dry night after a sunny day. It was perched on tangled vegetation 130 cm above the ground in dry shrubland besides recently cleared pasture. MZUA.RE0121 and MZUA.RE0122 were found actively moving at night between the branches 80-200 cm above the ground. ZSFQ D606 was found active during daytime after bulldozers opened a track in old-growth forest.
Distribution. Deciduous and semideciduous forests along the central Pacific coast in Ecuador in the provinces of Manabí and Guayas, at elevations between 5 and 317 m (Fig. 5).    Etymology. The specific name georgejetti honors George Jett, who has been a long-time donor to Rainforest Trust and has supported the reserves of Fundación Jocotoco in Ecuador. He is an international traveler with a passion for reptiles, amphibians, and birds.
Conservation status. We consider Dipsas georgejetti to be Vulnerable following the IUCN criteria A1c,B1a,b(iii, iv) (IUCN 2001) because its extent of occurrence is estimated to be 10,193 km 2 , it is known only from 9 localities effectively corresponding to 4 patches of forest lacking connectivity between them, and its habitat is severely fragmented and declining in extent and quality due to deforestation. At the type locality, D. georgejetti was found in a patch of deciduous forest of 13 km 2 that was being cleared to accommodate cattle pastures. One of the localities, 15 km N of Guayaquil, where D. georgejetti was collected in 1959, is now completely deforested, which suggests that this arboreal species is no longer present there.    Sibynomorphus oligozonatus Cadle, 2007: 195 (part).
Diagnosis. Dipsas oswaldobaezi is placed in the genus Dipsas based on phylogenetic evidence (Fig. 3) and the absence of a labial that is noticeably higher than other labials and in contact with the postocular, primary and secondary temporals. The species differs from all described species of Dipsas based on the following combination of characters: (1) 15/15/15 smooth dorsals with a slightly enlarged vertebral row (1-1.2 times as wide as adjacent rows); (2) loreal and prefrontal in contact with orbit; (3) six supralabials with 3 rd and 4 th contacting orbit; (4) no infralabials in contact behind symphysial; (5) 163-179 ventrals in males, 177-179 in females; (6) 68-70 divided subcaudals in males, 65-66 in females; (7) dorsal ground color light sandy brown with a pattern of 55-63 drab to brown black-edged middorsal blotches that are wider (7-9 vertebral scale rows) and solid down to the edges of the ventrals on the first one third of the body, but becoming narrower (1-3 vertebral scales long) and broken up laterally towards the tail; interspaces finely speckled with brown pigment; ground color of the head light sandy brown with a thin light cream nuchal collar and bold dark brown to black irregular blotches scattered on head plates and edging supralabials; ventral surfaces sandy brown with fine black speckling (Fig. 13b); iris sandy brown with dense dark brown speckling; (8) 277-348 mm SVL in males, 407-428 mm in females; (9) 85-114 mm TL in males, 110-122 mm in females.
Comparisons. Dipsas oswaldobaezi is most similar to D. williamsi, D. georgejetti, D. oligozonata, and D. vagrans, in that order, all of which were previously included in the genus Sibynomorphus. From D. williamsi, it differs in having 7-9 infralabials (vs. 10 in D. williamsi), first supralabial not in contact with prefrontal (vs. in broad contact in D. williamsi), and dorsal blotches that are lighter in the middle (vs. dark solid blotches). From D. georgejetti (Figs 11,12), it differs in having 6 supralabials with 3 rd and 4 th bordering the eye (vs. 7 supralabials with 4 th and 5 th bordering the eye in D. georgejetti). From D. oligozonata (Fig. 1o) and D. vagrans, it differs in having more than 160 ventrals. Dipsas oswaldobaezi further differs from D. oligozonata in having distinct bold crossbands at least middorsally along the whole length of the body, instead of being present only on the anterior half of the body. Genetic divergence in a 529 bp long fragment of the mitochondrial Cytb gene between D. oswaldobaezi and D. williamsi is 4.0-4.2%, whereas intraspecific distances are less than 0.2% in D. williamsi. For the same fragment, the distance between D. oswaldobaezi and D. georgejetti is 8.3%.
Natural history. Individuals of Dipsas oswaldobaezi have been found active by night on vegetation or at ground level in forested environments, pastures, or rural gardens. One individual (QCAZ 15108) was found hidden under leaf litter during daytime. Two individuals (MZUA.RE.0286 and QCAZ 14060) were found dead on roads.
Distribution. Deciduous and semideciduous lowland to lower montane forests and dry lowland shrublands in southwestern Ecuador (provinces of Loja and El Oro) and northwestern Peru (department of Tumbes), at elevation between 39 and 1289 m (Fig. 5).
Etymology. The specific name oswaldobaezi honors Dr. Oswaldo Báez, a renowned Ecuadorian biologist and researcher who has dedicated his life to the teaching of science, scientific thinking, and the conservation of nature. Oswaldo Báez has played a major role in science education in Ecuador through many popular science articles and books.
Conservation status. We consider Dipsas oswaldobaezi to be Vulnerable following the IUCN criteria B1a,b(iii, iv) (IUCN 2001) because its extent of occurrence is estimated to be 8,605 km 2 ; it is known only from eight localities effectively corresponding to four patches of forest lacking connectivity between them, and its habitat is severely fragmented and declining in extent and quality due to deforestation.
Remarks. In his revision of Dipsas oligozonata, Cadle (2007) allocated three additional specimens (AMNH 110587, BMNH 1935.11.3.108 and MUSM 2192) to a species known only from the holotype (EPN 3612), collected at Zhila, province of Azuay (S3.50280, W79.18808; 2795 m) (Fig. 5). AMNH 110587 was collected ca. 34 km airline distance from the type locality at an elevation of 2204 m, and it resembles the holotype in both color and lepidosis. However, BMNH 1935.11.3.108 and MUSM 2192 have more than 160 ventral scales and have broad dark brown crossbars that are at least twice as long as those present in both the holotype, AMNH 110587 and in the other four specimens of D. oligozonata examined by us (Table 2; Fig. 1o), all of which have fewer than 160 ventral scales and come from elevations between 2102 and 2891 m in the watershed of the Río Jubones (Fig. 5). The coloration and ventral scale counts in BMNH 1935.11.3.108 and MUSM 2192 are more similar to D. oswaldobaezi, and we designated them as paratypes of this species.

Systematics of the Dipsas peruana complex.
Based on differences in coloration and the topology of the molecular phylogeny obtained here (Fig. 3), we partition Dipsas peruana sensu Harvey and Embert (2008) into four allopatric species. This includes restriction of D. peruana to Peruvian-Bolivian populations, the resurrection of D. palmeri for populations ranging from northern Peru to central Ecuador, the description of a new species for northern Ecuador, and the resurrection of D. latifrontalis for populations in Colombia and Venezuela (Fig. 4).  Diagnosis. Dipsas klebbai is placed in the genus Dipsas based on phylogenetic evidence (Fig. 3), and the absence of a labial that is noticeably higher than other labials and in contact with the postocular, primary and secondary temporals. The species differs from all described species of Dipsas based on the following combination of characters: (1) 15/15/15 smooth dorsals with enlarged vertebral row (1.5-1.8 times as wide as adjacent rows); (2) one loreal and one preocular in contact with orbit; (3) 9-11 supralabials with (usually) 4 th to 6 th contacting orbit; (4) one pair of infralabials in contact behind symphysial; (5) 181-201 ventrals in males, 187-194 in females; (6) 99-123 divided subcaudals in males, 98-106 in females; (7) dorsal and ventral ground color light brown with various degrees of fine black speckling and 27-36 dark brown to black, cream-edged oblong blotches that are longer that interspaces and become smaller towards the tail (Fig. 2m, n); on first half of body, the dark bands meet ventrally to form full body rings; on second half they fail to meet ventrally; head black with different degrees of whitish edging on the labial scales, and a thin (1-2 scales long) cream to light brown irregular nuchal collar; dorsal blotches usually incomplete ventrally, extending far onto ventrals and occasionally fusing midventrally; cream edges of neighboring blotches fused in first 6-9 blotches; (8) 401-749 mm SVL in males, 525-630 mm in females; (9) 169-330 mm TL in males, 209-240 mm in females.

Dipsas klebbai
Comparisons. Dipsas klebbai is compared to species previously subsumed under D. peruana: D. latifrontalis, D. palmeri, and D. peruana. From D. latifrontalis (Fig. 1n) and D. palmeri (Figs 1r, s), it differs in having longer oblong to rectangular body blotches up to 7-13 vertebral scales long (vs. fewer than 8 vertebral scales long in D. latifrontalis and D. palmeri) that are also longer than the interspaces (Fig. 1l, m). Specimens of D. klebbai can be separated from specimens of D. peruana, with the exception of BMNH 1946.1.2078, based on the presence of the following characteristics (condition of D. peruana in parentheses): posterior body blotches twice to four times as long as interspaces (vs. posterior body blotches ca. equal in length or marginally longer than interspaces); interspaces never completely obscured by black pigment (vs. completely melanized in some specimens); dorsal surface of head black (vs. dark brown with dingy cream reticulations); dorsal body blotches fused ventrally on the first half of the body (vs. rarely fused); longest body blotch at least 7 vertebral scales long (vs. longest body blotch 4-7 vertebral scales long). Genetic divergence in a 684 bp long fragment of the mitochondrial Cytb gene between D. klebbai and D. palmeri is 8.2-9.2%, whereas intraspecific distances are less than 1.1% in both species. For the same fragment, the distance between D. klebbai and D. peruana is 10.7-11.0%.
Natural history. At night (21h53-02h13), specimens of Dipsas klebbai have been found active during or after light rain on arboreal vegetation 50-500 cm above the ground in a variety of environments ranging from primary montane cloud forests and evergreen montane forests to silvopastures and forest borders, occasionally close to rivers. By day, individuals have been found hidden underground in pastures or among shrubs in rural gardens, or coiled on leaves at 300 cm above the ground. At dusk, after warm days, individuals of Dipsas klebbai have been seen crossing roads. QCAZ 13124 laid six eggs on December 2014. Five eggs were found inside a rotten trunk at El Chaco, province of Napo Ecuador.
Distribution. Endemic to the eastern slopes of the Ecuadorian Andes in the provinces of Napo and Sucumbíos at elevations between 1246 and 2120 m (Fig. 4).
Etymology. Named after Casey Klebba, in recognition of his appreciation of and passion for Andean wildlife, and his invaluable support of AA's field expeditions to remote areas of Ecuador. After a visit to Peru in 2011, Casey became an active supporter of conservation and scientific projects in Ecuador.
Conservation status. All known localities of occurrence for Dipsas klebbai fall within the limits or within the buffer zone of the following protected areas: Parque Nacional Cayambe Coca, Parque Nacional Sumaco Napo Galeras, Reserva Ecológica Antisana, and Reserva Ecológica Cofán Bermejo. Furthermore, the species is common in degraded environments, which suggests a degree of tolerance for habitat modification. For these reasons, and because it does not meet the criteria (IUCN 2001) for qualifying in a threatened category, we here list it as Least Concern following IUCN guidelines.
Remarks. In their revision of Dipsas peruana, Harvey and Embert (2008) included specimens of D. klebbai. However, they found no characters that could diagnose these specimens from the rest of Ecuadorian and Peruvian specimens of D. "peruana" in order to establish species boundaries. They also grouped the then valid D. boettgeri, D. latifrontalis, and D. polylepis under D. peruana. The authors were right to point out that the different populations cannot be separated based on characters of lepidosis. However, they did not include molecular data in their analyses, and also failed to notice the geographically structured differences in the length of the body blotches and their relationship to the length of the interspaces.

Proposed standard English name. Palmer's Snail-Eater
Proposed standard Spanish name. Caracolera de Palmer Diagnosis. Dipsas palmeri differs from all described species of Dipsas based on the following combination of characters: (1) 15/15/15 smooth dorsals with enlarged vertebral row; (2) one loreal and one preocular in contact with orbit; (3) 8-10 supralabials with (usually) 4 th to 6 th contacting orbit; (4) one pair of infralabials in contact behind symphysial; (5) 172-202 ventrals in males, 181-200 in females; (6) 91-118 divided subcaudals in males, 86-102 in females; (7) dorsal and ventral ground color light brown with various degrees of fine black speckling and with 32-41 brown to blackish, white-edged circular blotches that are longer than interspaces in the first half of the body, but shorter in the second half (Figs 1r, s); adult head gray with different degrees of whitish edging on the labial scales, and a thin (1-2 scales long) white to light grayish brown irregular parietal collar; dorsal blotches incomplete ventrally, extending marginally onto ventrals but not fusing midventrally; (8) 215-907 mm SVL in males, 642-1187 mm in females; (9) 78-390 mm TL in males, 246-298 mm in females.
Comparisons. Dipsas palmeri is compared to species previously subsumed under D. peruana: D. latifrontalis, D. klebbai (Fig. 1l, m), and D. peruana. From D. latifrontalis (Fig. 1n), it differs in having the first 19-35 dorsal blotches edged with white or cream, vs. the first 9-10 in D. latifrontalis. The only known adult of D. latifrontalis photographed in life has bronze interspaces (Fig. 1n), a coloration not seen in any adult of D. palmeri. From D. klebbai, it differs in having shorter blotches (longest blotch up to 3-7 vertebral scales long) that are circular (instead of oblong) and that are only longer than the interspaces on the first half of the body. From D. peruana, it differs in having dorsal blotches that are shorter than interspaces on posterior half of the body, and in lacking melanized interspaces in adult individuals.
Distribution. Eastern slopes of the Ecuadorian and Peruvian Andes south of the Jatunyacu-Napo river valley in Ecuador and north of the Huancabamba depression at elevations between 1211 and 2282 m (Fig. 4).
Conservation status. An estimated 31 out of the 42 known localities of occurrence for Dipsas palmeri are located within the limits or the buffer area of the following protected areas: Bosque Protector del Alto Nangaritza, Parque Nacional Llanganates, Parque Nacional Podocarpus and Parque Nacional Sangay. Furthermore, the presence of the species in degraded environments suggests a degree of tolerance for habitat modification. For these reasons, and because it does not meet the criteria for qualifying in a threatened category, we here list it as Least Concern following IUCN guidelines.
Remarks. Neither Peters (1960) nor Harvey and Embert (2008) recognized the geographic morphological distinctiveness of Dipsas palmeri from Ecuador and Peru. Certainly, D. palmeri is most similar in coloration and lepidosis to D. latifrontalis (Fig.  1n) from Venezuela, and that is why Peters considered them synonyms. However neither Peters (1960) nor Harvey and Embert (2008) saw live specimens of D. latifrontalis in order to recognize the differences in life color pattern between the two species.
Two other junior synonyms of Dipsas peruana are D. latifasciata and D. polylepis, both of which occur in Peru (Fig. 4). Of these, only the latter must remain a synonym of D. peruana; the former should be transferred to the synonymy of D. palmeri, as defined here. Examination of photographs of the specimen of D. latifasciata (BMNH 1946(BMNH .1.2077) reveals this species has dorsal blotches shorter than interspaces on posterior half of the body, a character seen in D. palmeri but not in D. peruana. The holotype was collected by A. E. Pratt in "Upper Marañón", with no further specific locality mentioned. However, the type locality can be restricted to the immediate environs of the town of Jaén, as the "Upper Marañón" is considered the segment of the Marañón river that goes from the town of Jaén until the river meets the Santiago River. Additionally, in a letter to his wife in 1913, the explorer explains how he crossed the Ecuadorian Andes and arrived at the town of Jaén in northern Peru, where he stayed and collected specimens for the BMNH before proceeding to Iquitos along the Marañón river, with no mention of visiting any locality east of the river at elevations where D. palmeri and D. peruana are known to occur. Harvey and Embert (2008) pointed out that the Huancabamba depression could be a geographic barrier separating species within the D. peruana complex, but they did not find evidence to support this view. Our results suggest that the Huancabamba depression is a major geographic barrier separating D. palmeri (north) from D. peruana (south).
Comparisons. Dipsas peruana sensu stricto is compared to species previously subsumed under D. peruana sensu lato: D. latifrontalis, D. palmeri, and D. klebbai. From D. latifrontalis and D. palmeri, it differs in having dorsal blotches along the entire body similar in length or longer than interspaces (shorter than interspaces in D. latifrontalis and D. palmeri), and in having melanized interspaces in some adult individuals. With the exception of BMNH 1946.1.2078, specimens of D. peruana can be separated from specimens of D. klebbai by possessing at least one of the following characteristics: posterior body blotches similar in length or marginally longer than interspaces (twice to four times as long in D. klebbai); short circular to vertically elliptical body blotches usually only up to 4-7 vertebral scales long; melanized interspaces; dorsal surface of the head not completely black; and dorsal body blotches rarely fused ventrally.

Proposed standard English name. Broad-fronted Snail-Eater
Proposed standard Spanish name. Caracolera frentona Diagnosis. Dipsas latifrontalis differs from all described species of Dipsas based on the following combination of characters: (1) 15/15/15 smooth dorsals with moderately enlarged vertebral row; (2) one loreal and one preocular in contact with orbit; (3) 8-10 supralabials with 3 rd to 6 th contacting orbit; (4) one pair of infralabials in contact behind symphysial; (5) 192 ventrals in one male (CVULA 7883), 194 in the female holotype; (6) 109 divided subcaudals in the single male, 95 in the female holotype; (7) dorsal and ventral ground color bronze (light brown in juveniles) with 32-36 dark reddish brown to black, circular to vertically elliptical blotches that are longer than interspaces and white to cream edged on first half of body; head grayish brown to black with different degrees of whitish edging on the labial scales, and with or without a thin (1-2 scales long) dingy white irregular nuchal collar; dorsal blotches extending marginally onto ventrals and occasionally fusing on the anterior part of the body; (8) 800 mm SVL in the holotype female; (9) 220 mm TL in the holotype female.
Comparisons. Dipsas latifrontalis is compared to species previously subsumed under D. peruana: D. palmeri, D. peruana, and the herein described D. klebbai. From D. palmeri, it differs in having the first 9-10 dorsal blotches edged with white or cream, vs. the first 19-35 in D. palmeri. The only known adult of D. latifrontalis photographed in life has bronze interspaces (Fig. 1n), a coloration not seen in any adult of D. palmeri (see also Remarks below). From D. klebbai, it differs in having shorter blotches (longest blotch up to 6-8 vertebral scales long) that are circular (instead of oblong) and that are only longer than the interspaces on the first half of the body. From D. peruana, it differs in having dorsal blotches in posterior half of the body shorter than interspaces, and in lacking melanized interspaces in adult individuals.
Distribution. Known only from two localities in the Venezuelan Andes and one in the Northern Colombian Andes at elevations between 1000 and 1400 m (Fig. 4).
Remarks. Neither Peters (1960) nor Harvey and Embert (2008) examined the holotype of Dipsas latifrontalis, and they used Boulenger (1905) description to assign specimens of D. palmeri and D. peruana, respectively, to D. latifrontalis. We examined pictures of the holotype of D. latifrontalis from the BMNH, provided to us by César L. Barrio-Amorós. In coloration, the holotype is nearly identical to the uncollected adult presented in Figure 1n (San Isidro, Barinas province, Venezuela), with faint cream edging restricted to blotches 1-9, and indistinct blotches on the posterior part of the body. The previously only known photograph of a D. latifrontalis is of a juvenile from the same location as the specimen in Figure 1n (Rivas et al. 2012).

Discussion
Higher-level relationships within Dipsadini are still far from being resolved. The monotypic Plesiodipsas perijanensis was not included in our analysis or other recent molecular phylogenies. The species of Dipsas+Sibynomorphus and Sibon included here form monophyletic groups, but this is not the case for the genus Tropidodipsas, for which T. sartorii and T. fasciata + T. fischeri are the successive sister lineages of Dipsas+Sibynomorphus and Sibon (Fig. 3). This arrangement mirrors the results of Sheehy's (2012) unpublished PhD thesis, which presented evidence that groups consisting of T. sartorii, T. annulifera, T. fischeri, T. philippii, and T. fasciatus, as well as several new species of Tropidodipsas were not each other's closest relatives, and some merited recognition as distinct genera. Sheehy (2012) also presented phylogenetic evidence that Sibon sanniolus and Dipsas gaigeae do not belong to their nominal genera. Instead, each is more closely related to Tropidodipsas sensu stricto (D. gaigeae) or "T." sartorii + Geophis + "T." annulifera (S. sanniolus) than any species of Dipsas or Sibon.
Decades ago, Parker (1926) and Smith and Taylor (1945) suggested that Sibynomorphus and Dipsas were synonyms. More recently, Zaher et al. (2009), Grazziotin et al. (2012), and Sheehy (2012) recognized that Dipsas is paraphyletic with respect to Sibynomorphus, a conclusion we corroborate based on the results of our ML molecular phylogeny. In fact, members of former Sibynomorphus fall into four different clades across the phylogeny of Dipsas. In general, we suggest that the former Sibynomorphus species represent cases of convergent evolution; apparently from within several inde-pendent Dipsas clades or they represent an ancient morphotype successfully persisting through today.
Additionally, many traditional infrageneric groups are either non-monophyletic, or poorly supported and weakly placed. We recognize that this may reflect inadequate sampling of taxa (only 43 of 77 species are included) or characters (only four mtDNA and one nuclear locus were used). From the eight Dipsas species groups recognized by Harvey (2008) (Table 1), we only found phylogenetic support for the D. articulata and D. indica species groups. Two groups of species that are monophyletic in our molecular phylogeny and are similar in coloration and lepidosis are: 1) D. georgejetti + D. oligozonata + D. oswaldobaezi + D. williamsi, and 2) D. klebbai + D. palmeri + D. peruana. The sampled members of the D. oreas group are monophyletic if D. andiana is placed in this group, as it is the strongly supported (in both BI and ML analyses) sister taxon of D. oreas. We therefore place D. andiana in the D. oreas group and propose that the same be done for the morphologically similar D. nicholsi from Panama.
Dipsas bobridgelyi is most similar in coloration to D. gracilis (Fig. 1h, i). These species are recovered as sister taxa in our phylogenetic analyses (Fig. 3) and have nonoverlapping, but adjacent distribution ranges in western Ecuador (Fig. 4). This scenario suggests a parapatric speciation event, as the distribution of D. gracilis is congruent with Chocoan evergreen forest in northwestern Ecuador whereas the distribution of D. bobridgelyi is congruent with Tumbesian semi-deciduous forests in southwestern Ecuador.
Although we did not examine MUSM 17589 from Tumbes department, Peru, the description of the coloration and head scales of this specimen provided by Cadle (2005) and Harvey (2008) suggests that it is a Dipsas bobridgelyi, rather than a D. gracilis, as was originally suggested by both authors before the description of D. bobridgelyi herein. There is no other voucher of D. gracilis from Peru and it is unlikely that two morphologically and phylogenetically, and likely also ecologically very close species, occur in sympatry. Hence, from a biogeographic perspective, we suggest D. gracilis does not occur in Peru and that all specimens from south of the southern limit of D. gracilis in southwestern Ecuador and adjacent northwestern Peru represent D. bobridgelyi. Peters (1960) recognized a geographic morphological structure within the widely distributed Sibon nebulatus when he defined the subspecies nebulatus, leucomelas, hartwegi, and popayanensis. Here, our genetic results corroborate that S. nebulatus leucomelas from Ecuador and S. nebulatus hartwegi are distinct from the two Central American samples from Belize and northeastern Costa Rica, a divergence already put forward by Sheehy (2012). Yet, S. nebulatus is paraphyletic with respect to both S. dunni and S. bevridgelyi, which group with S. nebulatus leucomelas from Ecuador. Elevation of the two subspecies S. nebulatus leucomelas and S. nebulatus hartwegi to full species status would resolve this paraphyly. However, we refrain from taking this step because our sample size for S. nebulatus hartwegi is small, even though plenty of photographic data from references (e.g., Natera-Mumaw et al. 2015) and online sources confirm that long nuchal bands and often brownish color pattern are typical of S. nebulatus hartwegi occurring from Medellin, Colombia, east into Venezuela. In addition, the supposedly diagnostic darker ground color of S. nebulatus leucomelas with copious blackish stippling of the interspaces and head (Peters 1960) is not exclusive of this subspecies. There is ample evidence (photographic vouchers, preserved specimens, online photo sources) that this color pattern is rather consistent in S. n. nebulatus from Nicaragua through Panama, and can even be observed in single specimens as far as the northern limit of the species in Mexico. Furthermore, we have no genetic data of S. nebulatus from southern Costa Rica, Panama, and Colombia, which could confirm a clear split between two species, rather than a gradient of two intergrading subspecies.
Sibon bevridgelyi and S. nebulatus leucomelas were not recovered as sister taxa in our phylogenetic analyses (Fig. 3), despite being similar in coloration and lepidosis, and having adjacent marginally overlapping distribution ranges in western Ecuador (Fig.  8), a pattern that would suggest an allopatric speciation event. Our phylogeny suggests a more complex scenario that includes S. dunni from the dry valley of the Mira River in northwestern Ecuador. In any case, the three species are segregated geographically in western Ecuador, with S. n. leucomelas occupying the evergreen lowland and forest of northwestern Chocoan Ecuador, S. bevridgelyi the semi-deciduous forest in southwestern Ecuador, and S. dunni dry montane shrublands. Whether the current low genetic divergence between these three taxa constitutes a scenario of recent or ongoing gene flow between them is worth addressing further using nuclear markers. Strong local selection may have affected traits other than the mitochondrial genes.
Unlike the previous examples, the pattern of cladogenesis recovered in our phylogeny for the species of the Dipsas peruana complex (Fig. 3) suggest that a series of allopatric speciation events could be responsible for the current observed pattern of geographic genetic divergence between D. peruana and D. palmeri + D. klebbai. Two geographic barriers (i.e., Napo and Marañón rivers; Fig. 4) are located between the geographic ranges of the aforementioned species, and these features of the Andean geography have previously been recognized as important barriers to gene flow (Hackett 1993, Funk et al. 2007, Lynch Alfaro et al. 2015. A different scenario of speciation can be interpreted from the current distribution (Fig. 5) of the clade comprised by Dipsas georgejetti, D. oligozonata, D. oswaldobaezi, and D. williamsi. All of these species are adapted to dry shrublands, and the distribution of this vegetation type in northern Peru and south-central Ecuador is not continuous. We hypothesize that the discontinuity of dry shrubland west of the Andes in Ecuador and Peru is what explains best the observed pattern of geographic genetic divergence in this group of snakes.
We suspect that there are numerous additional species to be described across all genera of Dipsadini. Our results and the results of other recent researchers such as Sheehy (2012) indicate that additional taxonomic changes are also needed at the species-group and genus level to create a robust, stable taxonomy that agrees with the molecular phylogeny. Other morphological data such as visceral topology (e.g., Wallach 1995) suggest that morphological synapomorphies may exist for these clades, but are complex and difficult to identify accurately. Hence, in order to clarify species richness and higher-level to detailed relationships in Dipsadini, a systematically intensive revision that includes genetic, biogeographic, and morphological data from the greatest number of species representing the known genera is needed.

Appendix 1
GenBank accession numbers for loci and terminals of taxa and outgroups sampled in this study. Novel sequence data produced in this study are marked with an asterisk (*).