Systematics of Nothopsini (Serpentes, Dipsadidae), with a new species of Synophis from the Pacific Andean slopes of southwestern Ecuador

Abstract Within Dipsadinae, some recent authors have recognized a tribe Nothopsini containing the genera Diaphorolepis, Emmochliophis, Nothopsis, Synophis, and Xenopholis, on the basis of a number of putative morphological synapomorphies. However, molecular results suggest that Nothopsis, Synophis, and Xenopholis do not form a monophyletic group, while the remaining taxa are unsampled in recent molecular phylogenies. Here, DNA-sequence data for some Diaphorolepis and Synophis species are provided for the first time, as well as additional new sequences for Nothopsis and some Synophis species. Including these and other existing data for nothopsine species, previous studies showing that Nothopsini is not a natural group are corroborated. Nothopsini Cope, 1871 is restricted to Nothopsis. Diaphorolepidini Jenner, 1981 is resurrected and re-delimited to include only Diaphorolepis, Emmochliophis, and Synophis. Finally, Xenopholis remains Dipsadinae incertae sedis. Known material of Diaphorolepidini is reviewed to generate revised and expanded descriptions and diagnoses at the tribe, genus, and species level. Numerous cryptic species are likely present in Synophis bicolor and Synophis lasallei. Finally, a new population from the low-elevation cloud forests of SW Ecuador is reported upon, which is genetically and morphologically distinct from all other species, that is here named Synophis zaheri sp. n.

Thus, the tribe Nothopsini does not appear to represent a natural group, despite the putative morphological synapomorphies uniting the taxa listed above (Savitzky 1974;Ferrarezzi 1994;Wallach 1995;Martinez 2011). Contrastingly, the strength of the molecular results suggests that these likely represent convergence, at least between Nothopsis and Xenopholis. This is not surprising, given the massive ecomorphological diversification exhibited by Dipsadinae following their adaptive radiation in the Neotropics (Cadle 1984a, b, c).
However, Diaphorolepis and Emmochliophis have still not been sampled in any molecular phylogeny, and it is thus unclear where their phylogenetic affinities lie. Morphological evidence suggests that these two genera form a clade with Synophis (see Hillis 1990). Furthermore, there are multiple species of Synophis, with potentially unclear species boundaries (Bogert 1964;Fritts and Smith 1969;Sheil 1998;Sheil and Grant 2001). Here, we report on new material from Diaphorolepis, Synophis, and Nothopsis, present a new molecular phylogeny, and describe a new species of Synophis. We review current knowledge of Diaphorolepis, Emmochliophis, and Synophis, and discuss species limits in these genera. Dipsadine diversity in the Andes is clearly underestimated, and new species are still being discovered in the 21st century (e.g., Salazar-Venezuela et al. 2014;Sheehy et al. 2014;Zaher et al. 2014).
We set up PCR reactions to a total volume of 25 µL containing MgCl2 2-3 mM , dNTPs 200 µM, 0.2 µM of each primer (0.8 µM in the case of ND4) and 1.25 U (16S and Cytb) or 0.625 U (ND4 and c-mos) of Taq DNA polymerase (Invitrogen). Thermocycling parameters consisted of an initial three-minute step at 94 °C; 25 to 30 cycles of 45-60 sec at 94 °C, 45 (16S and c-mos) or 60 (ND4 and Cytb) sec at 53-60 °C, 1 (16S and c-mos) or 2 (ND4 and Cytb) min at 72 °C; and a final extension of 7 min at 72 °C. We used 1.5% agarose gels to visualize the PCR products and QIAquick PCR purification Kit (QIAGEN) to remove unincorporated primers and dNTPs from every PCR reaction before they were sent to Macrogen Inc. for sequencing.
We combined these new data with the publically available sequences for Nothopsis and Xenopholis (Vidal et al. 2010;Grazziotin et al. 2012). We obtained additional sequences of S. bicolor from the Museu de Zoologia da Universidade de São Paulo (MHUA 14577 [Museo de Herpetología de la Universidad de Antioquia], from Colombia: 12S, 16S, CYTB, and CMOS) and the University of Texas, Arlington (UTA-R 55956 from Ecuador: CYTB and ND4).
We then included all publically available dipsadine species sampled for these genes. This matrix contains 24% missing data ('-'), but these have been shown not to have deleterious effects on taxon placement and support in previous analyses (e.g., Pyron  (Katoh and Standley 2013) under the default parameters in Geneious 7.1.9 (Biomatters Ltd.). We determined the optimal partitioning strategy using PartitionFinder (Lanfear et al. 2012). We estimated the phylogeny using MrBayes 3.2.5 (Ronquist et al. 2012), with 4 runs of 4 chains each, run for 20 million generations with the first 25% discarded as burnin. Convergence was assumed as the average standard deviation of split frequencies went to zero and the potential scale reduction factors went to one (Ronquist et al. 2012). The GenBank accession numbers for the new and existing data are given in Appendix I.

Morphological data
Species in Diaphorolepis, Emmochliophis, and Synophis have traditionally been delimited using easily determined external morphological characters (Bogert 1964;Hillis 1990). We relied here on a set of these characters, scored for museum specimens and our new material, to examine and delimit species boundaries (Table 1). For available specimens examined in person, in photographs, or in the literature, we recorded SVL and TL in mm, and counts of supralabials, infralabials, postoculars, ventrals, and subcaudals. We made cursory notes on the hemipenes of some male specimens when they were visible (Zaher 1999;Martinez 2011).

Molecular phylogeny
The overall topology and support (Figs 1, 2) is similar to numerous recent studies (Zaher et al. 2009;Vidal et al. 2010;Pyron et al. 2011;Grazziotin et al. 2012). We consider strong support to be posterior probabilities ≥95%, following recent authors (Felsenstein 2004). Overall, there is low support for many backbone nodes, which may reflect inadequate sampling of taxa (only ~250 out of ~900 dipsadine species) or characters (only two independent loci). Species in Dipsadinae can be broadly grouped into a primarily North American clade (Contia to Carphophis when viewing Fig. 1), a primarily Central American clade (Diaphorolepis to Atractus in Fig. 1), and a primarily South American clade (Crisantophis to Apostolepis in Fig. 2), though many species in the latter two clades range across both Central and South America. Several speciose genera in the primarily Central American clade are non-monophyletic, including Imantodes, Hypsiglena, Geophis, Sibon, Dipsas, Sibynomorphus ( Fig. 1), as in previous studies (Grazziotin et al. 2012;Pyron et al. 2013).
In agreement with previous results (Grazziotin et al. 2012;Pyron et al. 2013), we find that Nothopsini is not a natural group (Fig. 1). The genus Nothopsis is strongly supported, and strongly placed with Leptodeira + Imantodes within the Central American clade. Correspondingly, Xenopholis is strongly supported and weakly nested within the South American clade, as the sister lineage to Hydrodynastes. It appears that one Xenopholis scalaris (KU 222204) from a previous study (Pyron et al. 2011) may have been misidentified, and is actually related to X. undulatus. This specimen is strongly supported as the sister lineage to the sampled X. undulatus (R-6955), to the exclusion of the three other sampled X. scalaris, which are strongly supported as a monophyletic group. This specimen is from the Peruvian Amazon and is pictured in Duellman and Mendelson (1995). The specimen pictured resembles the Amazonian X. scalaris, rather than the more xeric X. undulatus from the Brazilian shield. Thus, it is possible either that a curatorial or laboratory error occurred at some point, or that there is cryptic genetic diversity in Xenopholis.
A strongly-supported clade comprising Diaphorolepis and Synophis represents the sister to the large, primarily Central American clade that also contains Nothopsis. Monophyly of Synophis with respect to Diaphorolepis is weakly supported. Within a weakly paraphyletic S. bicolor, there are three deeply divergent lineages, and the sampled specimen of S. lasallei. An apparently new species of Synophis is the stronglysupported sister lineage of S. calamitus. The species S. plectovertebralis remains unsampled in the molecular phylogeny. Although Emmochliophis is not sampled, we follow previous authors in assuming a close relationship with Diaphorolepis and Synophis, given their strong resemblance (Savitzky 1974;Hillis 1990). Thus, the synapomorphies previously used to diagnose Nothopsini (Savitzky 1974;Wallach 1995) apparently represent convergence in at least three distantly related dipsadine lineages.

Systematics
We seek here to only name clades associated Nothopsini that are strongly supported in our molecular phylogeny. Above the genus level, Nothopsini is not a natural group in any of its recent conformations. We place Nothopsis alone in Nothopsini Cope, 1871. We resurrect and re-delimit the tribe Diaphorolepidini Jenner, 1981 to include only Diaphorolepis, Emmochliophis, and Synophis. The genus Xenopholis is not strongly supported in any supra-generic group and remains incertae sedis in Dipsadinae (see Grazziotin et al. 2012). Our molecular and morphological data (Tables 1-3; Figs 1, 2) also corroborate previous authors in finding that genus and species boundaries within Diaphorolepidini are unclear and in need of revision (Sheil and Grant 2001). We here provide photographs and range maps of representative material (Figs 3-9). A number of issues are immediately apparent, and can be addressed with our results. We outline these below.
First, the head scalation of Diaphorolepis wagneri has not been accurately characterized by most authors (see Bogert 1964). Additionally, the holotype of D. laevis was incorrectly described with respect to several major characters (Werner 1923). Finally, reviewing museum specimens, including most holotypes, reveals that the current species boundaries and diagnoses are oftentimes inaccurate with respect to the observed range of variation in the relevant characters. In particular, the holotype of S. bicolor does not match many populations typically referred to this species (Bogert 1964;Hillis 1990;Sheil and Grant 2001).
In the case of Diaphorolepis wagneri, the postoculars can range from 1-3 (rather than 1-2), as illustrated by Bogert (1964), but not discussed explicitly. Werner (1901) apparently considered the small, lower postocular to be a subocular. Occasionally, the middle postocular will not be in contact with the brille, and resembles a temporal, behind the two remaining postoculars. As noted previously, the nasals are never divided, but only creased (Sheil and Grant 2001), contrary to reports from some previous authors (Bogert 1964;Hillis 1990).
In the case of Diaphorolepis laevis, Werner (1923) diagnosed the species as having fewer ventrals and subcaudals than D. wagneri, and smooth dorsal scales. Examination of the holotype (NMW 14860) reveals that it is indeed keeled, albeit weakly, throughout most of the midbody and posterior dorsal scale rows. This includes a bicarinate vertebral scale row that was previously considered to be diagnostic only of D. wagneri. The specimen appears to have a lighter-colored nuchal collar, though this may be a preservation artifact. The type locality within Colombia is unknown.
In the case of Synophis bicolor, the holotype (MZUT 257) has 180 ventrals, 136 subcaudals, and 9 infralabials, whereas sampled populations from the Andes of Ecuador typically have 152-166 ventrals, 96-122 subcaudals, and 10 or 11 infralabials. The locality of the holotype is unknown. Sampled populations from the Chocó of Ecuador match the holotype more closely, with 174-183 ventrals, 129-143 subcaudals, and 9-11 infralabials. The Chocóan populations typically occur at low to middle elevations (~200-300m), whereas Andean populations occur at higher elevations (~800-1700m). Populations from the northern western Andes of Colombia have 184-193 ventrals, 127-131 subcaudals, and 10-12 infralabials. These three populations (Chocóan, Colombian Andean, and Ecuadorean Andean; Figs 3D, 4), correspond to three deeply divergent genetic lineages within Synophis bicolor (Fig. 1). A full revision of this species complex is pending further molecular and morphological sampling. We refer to the Chocóan populations as S. bicolor, the Ecuadorean Andean populations as S. aff. bicolor, and the Colombian Andean populations as S. cf. bicolor (using aff. versus cf. somewhat arbitrarily) for the remainder of the paper. The S. bicolor group is also weakly paraphyletic with respect to the sampled specimen of S. lasallei, which is the sister lineage of the Ecuadorean Andean lineages. The specimen of S. lasallei (MZUTI 4181) strongly matches the other S. lasallei specimens examined (Table 1), and is thus not a mis-identified S. bicolor.
Finally, we report here on two specimens of Synophis aff. calamitus from low to middle elevations on the Pacific versant of the Andes in SW Ecuador. These are diagnosable from the species above based on numerous characters, and we here name them: Paratype. MZUTI 3355 (Fig. 3B), adult male collected a few minutes after the holotype, a few meters away.
Etymology. Named after the preeminent Brazilian herpetologist Hussam El-Dine Zaher, for his innumerable contributions to South American herpetology and snake systematics.
Description. Small-sized snakes (351-372mm SVL, 184-194mm TL) with slender bodies and head distinct from neck. Eye large (>1/3 head height), bulbous, and black in life, with pupil not easily distinguishable from iris. Pupil round in preservative (though this may be an effect of fixation). Dorsum coloration grayish-brown with iridescent sheen in life and preservation, no light-colored nuchal collar in adults, and posterior supralabials mostly pigmented (>50%). Ventral coloration primarily bright yellowish-white, extending onto margins of ventral scales and supralabials. Posterior one-third of ventral surface anterior to vent becomes increasingly mottled, and ventral surface of tail color of dorsum. Squamation pattern includes 166-169 ventral scales, 111-112 subcaudals, 19-19-17 dorsal scale rows (scale-row reduction of 2 rows past midbody), anal single, no apical pits, mid-body dorsal scales with weak single keel (first few dorsal scale-rows smooth), vertebral scale row not enlarged, nuchal scales smooth, 8 supralabials , 8 or 9 infralabials, 2 postoculars, loreal present, nasal undivided, fused prefrontals, internasals in contact, and rostral concave. Condition of the vertebrae, which are heavily modified in Emmochliophis and Synophis (Fritts and Smith 1969;Savitzky 1974;Hillis 1990) unknown, pending skeletal preparation or micro-CT scanning. Everted hemipenes are slightly bilobed, semicalyculate, and semicapitate, relatively stout and bulbous, covered in large spines or hooks, similar to that of Diaphorolepis and Synophis aff. bicolor and S. lasallei (Bogert 1964;Zaher 1999;Martinez 2011). Both specimens were active by night in primary evergreen foothill forest, with canopy cover between 70 and 100%. The holotype MZUTI 3353 was found on the ground, whereas the paratype MZUTI 3355 was found 50 cm above the ground in a bush. Neither were found close to water, but were active after a rainy day.
In light of this new species and the updated material we have located and examined (Tables 1, 2), we have prepared updated accounts for the tribe and the other species. Hopefully, these will serve as useful descriptive summaries for taxonomic boundaries, species delimitation, and the assignment of new specimens and populations to species-level groups. We focus primarily on the external morphological characters that will be of greatest use for identifying specimens in the field and from preserved collections. In some cases, more detailed information can be found in the original descriptions cited. The tribe name Diaphorolepidini was introduced in the PhD thesis of Jenner (1981), for which availability as a published work is ambiguous. We conservatively continue to credit the name to her, rather than treat it as unavailable and re-describe it ourselves.

Tribe Diaphorolepidini Jenner, 1981
Diaphorolepis Jan, 1863 (type genus by original designation) Emmochliophis Fritts & Smith, 1969Synophis Peracca, 1896 Etymology. Apparently from the Greek diaphoros for "differentiated" and lepis for "scales," likely referring to the enlarged vertebral scale row as compared to the rest of the dorsal scales.

Description. A group of relatively small-sized (<550mm SVL) dipsadine snakes restricted to the Darien of Panama and northern Andes of South America with fused prefrontals and either an expanded vertebral scale row (Diaphorolepis) or expanded zygapophyses and neural spines in adults (Emmochliophis and Synophis).
Notes. The tribe name has also been spelled 'Diaphorolepini' by Sheehy (2012), but Diaphorolepidini is the correct spelling based on the suffix -lepis, for which the stem is -lepid + -ini. This is a greatly restricted definition of Diaphorolepidini over the original description (Jenner 1981), which included Atractus, Chersodromus, Crisantophis, Elapomorphus, Enulius, Gomesophis, Pseudotomodon, Ptychophis, and Sordellina.

Genus Diaphorolepis Jan, 1863
Diaphorolepis laevis Werner, 1923 Diaphorolepis wagneri Jan, 1863 (type species by monotypy) Etymology. Apparently from the Greek diaphoros for "differentiated" and lepis for "scales," likely referring to the enlarged vertebral scale row as compared to the rest of the dorsal scales.
Notes. This genus was validly described by Jan (1863), and re-described by Werner (1897). Werner (1901) later incorrectly deemed Jan's name a nomen nudum, and re-described the genus and type species, designating a neotype. However, this was an error of interpretation, later realized by Werner himself (Werner 1929), and neither the re-description or neotype designation have any nomenclatural validity (see Bogert 1964). The lower subcaudal counts for some specimens likely represent truncated tails. Werner, 1923 Holotype. NMW 14860, locality given only as "Colombia."

Diaphorolepis laevis
Etymology. Apparently from the Latin laevis for "smooth," referring to the anterior dorsal scales.
Notes. Known only from the type specimen. The original description states that the dorsal scales are smooth, but weak keels are evident throughout the posterior portion of the body. A specimen at Harvard, reportedly from Leticia, Amazonas, Colombia, bears the identification Diaphorolepis laevis (MCZ R-143839). Upon examination, this specimen is clearly not Diaphorolepis on the basis of divided prefrontals (versus united in Diaphorolepis), lack of an enlarged bicarinate vertebral scale row (versus presence), and presence of an ocellated dorsal color-pattern (versus uniformly colored dorsum). The overall resemblance is of Dipsas sp.
Etymology. Most likely after Moritz Wagner, who collected the holotype (see Bauer 2013), and not Johann Andreas Wagner as suggested by previous authors (Beolens et al. 2011).
Description. Relatively small-sized (~250mm SVL) terrestrial snakes restricted to the Pacific Andean slopes of NW Ecuador, with a small number (<17) of maxillary teeth, 8 supralabials, 8 infralabials, fused prefrontals, internasals in contact, loreal absent, fewer than 150 ventrals, fewer than 100 subcaudals, dorsal scales in 19 rows without reduction, trunk vertebrae with lateral expansion of the zygapophyses, and expanded zygapophyses forming a rod-and-groove mechanism in Emmochliophis fugleri, but not in E. miops.
Notes. Both species are known only from the types. The hemipenis of E. fugleri has been briefly described (Fritts and Smith 1969), but prior to modern classifications of the organ (Zaher 1999), and needs to be examined in more detail. The organ is unknown in E. miops, as the sole known specimen is female (Sheil 1998).
Etymology. After Dr. Charles Fugler, who collected the holotype. Description. A terrestrial snake from the Pacific Andean slopes of NW Ecuador, diagnosable by 16 maxillary teeth, 8 infralabials, 8 supralabials, 2 postoculars, internasals in contact, loreal absent, nuchal collar absent, 140 ventrals, 97 subcaudals, dorsal scales in 19 rows without reduction, strong keels, and zygapophyses expanded laterally forming rod-and-bar assembly. Type locality is surrounded by banana plantations. Little else is known about the habits or habitat of the species.
Notes. Known only from the type specimen, a male, collected by C. Fugler in February 1966.
First are the Ecuadorean Andean highlands populations (Synophis aff. bicolor), which occur both on both the Pacific and Andean versants (~800-1700m). These are diagnosable by number of ventrals (152-166), subcaudals (96-122), infralabials (10 or 11), and supralabials (8 or 9), in combination. One individual (UMMZ 91550) has 24/27 maxillary teeth. The southernmost individual we examined (MZUTI 4180) has a very low number of ventral scales (152) compared to the remaining populations (160-166). Populations east and west of the Andes may also be a distinct species (O. Torres-Carvajal, pers. comm.), and are presented separately here. Most records from the Pacific versant north of the Río Toachi appear to represent S. calamitus (see below); one specimen reported from north of the river (BMNH 1940.2.30.31) may be mislabeled, mis-identified, or the locality mis-referenced, or the species may be sympatric at some localities north of the river.

Notes.
A detailed description was also provided by Hillis (1990). The hemipenes have likely not been examined. Easily confused with Synophis bicolor; at least one specimen (QCAZ 11931) from near the type locality was originally mis-identified (O. Torres-Carvajal, pers. comm.). We suggest that all populations north of the Río Toachi are likely to represent S. calamitus. As mentioned above, one specimen apparently matching S. bicolor (BMNH 1940.2.30.31) is known from Río Soloya near Mindo north of Río Toachi, but this may have been mis-labeled, or mis-referenced geographically. The specimen of "S. bicolor" examined by Zaher (1999), QCAZ 452, cannot be located (O. Torres-Carvajal, pers. comm.), but originates from Chiriboga, Pichincha Province, Ecuador, north of Río Toachi, and thus may represent an S. calamitus. If this is the case, the hemipenes of S. calamitus and S. lasallei are nearly identical (Zaher 1999;Martinez 2011). Finally, one specimen sequenced here from Tambo Tanda (MZUTI 3694) appears to have aberrantly subdivided head scales, possessing one extra postocular, and 2 extra supralabials and infralabials (Fig. 8), which are misshapen and abnormally small. The badly damaged paratype also appears to have two postoculars on one side (O. Torres-Carvajal, pers. comm.). Thus, we concur with Hillis (1990) that one postocular, 7 or 8 supralabials, and 9 infralabials (along with the divided internasals and smooth anterior dorsal scale-rows) are generally diagnostic of the species, but with rare individual variation.
Notes. The hemipenes are very similar to both Diaphorolepis and S. bicolor (Bogert 1964;Zaher 1999;Martinez 2011). Much like Synophis bicolor, this species as currently described has a large geographic and elevational range, with wide variation in phenotype. There is significant variation in the number of dorsal scale rows and reduction thereof. One specimen from Ecuador (MCZ R-156873) has only one postocular and 7 supralabials, but otherwise matches the species. All other specimens have 2 and 8, respectively. Another specimen from Ecuador (MECN 2220) has 165 ventrals and 117 subcaudals with 19-19-17 scale rows, and is thus indistinguishable from S. aff. bicolor, with the exception of the strong keels on the nuchal scales and geographic distance from the nearest highland populations of S. aff. bicolor. All other specimens of S. lasallei have 144-156 ventrals, and most have (21-23)-(21-22)-(19-21) dorsal scale rows. Thus, it seems exceptionally likely that this is a species complex, possibly divided between highland and lowland, or northern and southern populations.
Paratype. UVC 11580, from type locality. Etymology. From the Latin plecto-for "braided" or "woven" and veretbralis for "vertebrae," referring to the appearance of the interlocking zygapophyses viewed from above (Sheil and Grant 2001).
Notes. Known only from the holotype and paratype (apparently juveniles), though other material has apparently been collected in Colombia, near the type locality (T. Grant and E. Meneses-Pelayo, pers. comm.). The hemipenes have not been examined. A more detailed description of the two specimens is provided by Sheil and Grant (2001).
Given our restriction of the name, we also provide the following re-description of the re-delimited Nothopsini. Note that we have not performed a comparative examination of a large series of preserved material, and these data are summarized from the literature (Dunn and Dowling 1957;Savage 2002;Kohler 2008;McCranie 2011) to provide a basis for future revisions.
Description. A relatively small-sized (<350mm SVL) dipsadine snake, ranging in Central and South America from Honduras to Colombia and Ecuador, in lowland and middle-elevation rainforests, 250-900m, distinguishable from nearly all other similar or related snakes in the area by the rugose, granular nature of the dorsal scales, in particular lacking differentiation of the cephalic scales with the exception of well-defined internasals and poorly defined frontal and parietals, which are separated by rows of irregular, undifferentiated scales. Color pattern consists of irregular and poorly defined blotches of blackish or light, dark, and yellowish brown. With respect to the characters described here for diaphorolepidine species, Nothopsis rugosus typically exhibits 19-21 maxillary teeth, 9-13 supralabials, 11-16 infralabials, 149-162 ventrals, 81-112 subcaudals, dorsal scales in (24-30)-(26-30)-(22-26) rows, SVL of 151-320mm, and tail length of 61-133mm (see Dunn and Dowling 1957).
Notes. This taxon has historically been divided up into as many as three species (see Dunn and Dowling 1957), though only a single species is currently recognized. There may be cryptic variation or undiscovered diversity within this group. Note that the family name was originally spelled Nothopidae by Cope (1871), but -ops-is the correct stem from -opsis, and Nothopsidae (and Nothopsini) is thus the correct spelling, as adopted by later authors.

Systematics of Diaphorolepidini and Nothopsini
Corroborating previous results, we find that current supra-generic classification in Dipsadinae does not accurately reflect the phylogeny and describe natural groups in many cases (Pyron et al. 2011;Grazziotin et al. 2012). Support for monophyly and placement of many genera is low, and many other genera are apparently non-monophyletic. Efforts to clarify this situation are underway, sampling more taxa and characters (F. Grazziotin, pers. comm.). Only ~250 out of ~900 dipsadine species (Wallach et al. 2014) are sampled here for a few genes, but cryptic and undiscovered diversity is likely much higher in the group, and will require extensive additional sampling of taxa and characters to arrive at a stable phylogenetic and taxonomic resolution. The taxonomy of Dipsadinae has been contentious for quite some time (Cadle 1984a,b,c;Zaher 1999;Zaher et al. 2009;Grazziotin et al. 2012;Sheehy 2012), and will likely require extensive additional sampling of taxa and characters to provide a stable taxonomic resolution.
In particular, we find that Nothopsini is not monophyletic as historically defined, but that Nothopsis is strongly nested within a primarily Central American clade, with Imantodes and Leptodeira. We restrict tribe Nothopsini Cope, 1871 to Nothopsis. We resurrect and re-delimit Diaphorolepidini Jenner, 1981 to include only Diaphorolepis, Emmochliophis, and Synophis. Whereas Emmochliophis remains unsampled in the molecular phylogeny, it appears to be the sister-taxon of Synophis based on morphological data (Hillis 1990). However, our phylogeny suggests that many of the morphological characters previously used to define supra-generic groups in Dipsadinae (see Savitzky 1974;Wallach 1995) are subject to strong and rapid convergence. Thus, future studies may find an alternative placement for this genus. Finally, the genus Xenopholis is weakly nested within a primarily South American clade, and remains Dipsadinae incertae sedis.

Species limits in Diaphorolepidini
Larger sample sizes reveal expanded ranges of diagnostic characters previously used to delimit species in Diaphorolepidini. These will hopefully assist future researchers in describing new taxa, and re-delimiting species boundaries. In particular, both Synophis bicolor and S. lasallaei may comprise multiple distinct species. Additional DNA sequencing and meristic and mensural measurements of more specimens should help clarify taxonomic boundaries.
In the case of Synophis bicolor, the Chocóan populations in Ecuador and presumably nearby Colombia match the description of the holotype, and thus likely represent the source of the original specimen, which remains to be re-described in detail. Contrastingly, highland populations in the Andean Highlands of Ecuador and Colombia are morphologically and genetically distinct, and both likely represent undescribed species. In the Ecuadorean Andes, populations of this taxon occur on both the Pacific and Amazonian versants, which may also be distinct from each other. The sampled specimen of S. lasallei is weakly nested within the sampled specimens of S. bicolor. A wide range of squamation and color pattern is observed in S. lasallei, which may represent cryptic species, as well as potential mis-identification of examined specimens. Finally, a cloud-forest population from the Pacific versant in SW Ecuador represents a new species described here as S. zaheri, allied to S. calamitus. Understanding the geographic distribution and genetic diversity in these taxa will require additional genetic sampling, which is hampered by the rarity of these species.
One of the most distinctive features of diaphorolepidine species is the highly modified condition of the vertebrae, in which the prezygapophyses and postzygapophyses are broadly expanded, forming ridges, and occasionally interlocking (Bogert 1964;Fritts and Smith 1969;Hillis 1990). Given the difficulty of preparing the skeletal material and the extreme rarity of specimens, this was not examined for S. zaheri or any additional specimens examined here. However, this may be a crucial character for future systematic revisions in the group, possibly utilizing micro-CT scanning or radiography.
Another possible source of information for delimiting species are the hemipenes. The organs are highly similar in Diaphorolepis and most Synophis species (Bogert 1964;Jenner 1981;Hillis 1990;Zaher 1999). Our observations agree with previous authors that the hemipenes are not strongly differentiated among species, though larger comparative series may reveal characters that serve to better diagnose species-level groups. In particular, the hemipenes are "nearly identical" in S. bicolor and S. lasallei (Zaher 1999;Martinez 2011), and our examination of S. zaheri shows no obvious qualitative differences. It is possible that speciation is primarily ecological or allopatric in this group, and thus there is little physical reproductive isolation.

Conclusions
Higher-level taxonomy in Dipsadinae is still partially unresolved, and many genera and supra-generic groups are either non-monophyletic, or poorly supported and weakly placed. This includes Nothopsini Cope, 1871, which must be restricted to Nothopsis, if it is used at all. We resurrect and re-delimit Diaphorolepidini Jenner, 1981 to include only Diaphorolepis, Emmochliophis, and Synophis. The genus Xenopholis remains Dipsadinae incertae sedis. Revised and expanded diagnoses in Diaphorolepidini support the distinctiveness of all currently recognized taxa. Cryptic species are likely present in S. bicolor and S. lasallei. A new population from the cloud forest of SW Ecuador is morphologically and genetically distinct, and we here name it S. zaheri. We hope that these data will provide a robust platform for future researchers to examine species boundaries in Diaphorolepidini, as additional work clearly remains to be done. This is hampered, however, by the extreme rarity of these species.