Molecular phylogeny and taxonomy of the Epictia goudotii Species complex (Serpentes: Leptotyphlopidae: Epictinae) in Middle America and northern South America

Here we review the systematics of the threadsnakes of the Epictia goudotii Species complex in Middle and northern South America using external morphology and molecular data. Two species, Epictia goudotii and E. magnamaculata, are currently recognized from that region, but we provide evidence for recognizing, as species, three other nominal forms usually treated as subspecies of E. goudotii: E. ater, E. bakewelli, and E. phenops. Thus, together with E. columbi (Bahamas), we recognize six species in the Epictia goudotii Species complex. Because E. albifrons from northern South America has been confused with E. goudotii in the past, we also briefly discuss the taxonomic status of that species and its apparent close relative E. tenella, which are not members of the E. goudotii complex.


INTRODUCTION
The Genus Epictia, Gray (1845) contains approximately 25 species of threadsnakes occurring in the New World, although nearly all of them are restricted to South America (Adalsteinsson et al., 2009). One wide-ranging species, Epictia goudotii (Duméril & Bibron, 1844;as Stenostoma Wagler, 1824), occurs in northern South America and Middle America and has had a long taxonomic history during which some populations have been recognized as either subspecies or species. Adalsteinsson et al. (2009) provided molecular evidence that Epictia goudotii is actually a complex of species and elevated one previously recognized subspecies to a species (E. magnamaculata (Taylor, 1940;as Leptotyphlops)). Those results of Adalsteinsson et al. (2009), along with those of McCranie (2011) noting that some of the morphological variation reported in the literature for E. goudotii was also associated with geography, prompted this current study. Herein, we gather new molecular and morphological evidence that justifies the recognition of additional species of snakes in this Species complex.

MATERIALS AND METHODS
The populations of Epictia goudotii under study herein are those occurring on the mainland from Mexico to northern South America, and those on the Islas de la Bahía and Swan Islands of Honduras. We also include the Bahamian species E. columbi (Klauber, 1939) shown to be closely related to E. magnamaculata in a previous molecular analysis ( Adalsteinsson et al., 2009). Because of the unavailability of genetic material from much of South America, we have focused primarily on populations in Middle America (= Mexico and Central America) for this current study. The type locality of E. goudotii lies in Colombia, but unfortunately Colombian authorities do not allow exportation of tissued material from that country. Specimens were collected and exported by JRM with permission of the Honduras government, under permits Resolució n DE-MP-102-2012 and Constancia 036-2012-DVS-ICF, Constancia 038-2012-DVS-ICF.
Alignments were performed with (MUSCLE) in MEGA 6.06 (Tamura et al., 2013). The total concatenated alignment for the two genes was 1,269 aligned sites: 12S-471 sites, cytb-798 sites. A maximum likelihood analysis was performed using RAxML 8.0.9 (Stamatakis, 2014) through the CIPRES Science Gateway (Miller, Pfeiffer & Schwartz, 2010). The data were divided into four partitions (12S, cytb by codon position), and were analyzed using the evolutionary model GTRGAMMA, the maximized available option in RAxML. The same partitioning scheme was used for these genes in the study of Adalsteinsson et al. (2009), which involved sequences of the same species and where different partitioning schemes resulted in nearly identical trees. Gaps were treated as missing data. All parameters for the ML analyses were estimated by the program during the run. Branch support in the trees was provided by bootstrap analysis (2,000 replicates).
Authors of all literature mentioning dorsal head scales regarding Epictia goudotii and relatives from Mexico, Central America, and northern South America had called the dorsal head scale following the rostral scale a prefrontal scale, with that prefrontal scale fused with the rostral in some populations (i.e., Savage, 2002). That was until Pinto et al. (2010) called that prefrontal scale a frontal for their descriptions of E. goudotii and E. magnamaculata in Colombian territory. The source of the Pinto et al. different terminology was not cited, but came from Wallach (2003). Since the new terminology is likely to be confusing to workers familiar with the previous literature (all literature on E. goudotii complex except Pinto et al., 2010), we include an illustration of the head of two specimens with the new terminology illustrated (Fig. 1) and the old terminology explained. We use this new terminology knowing it will likely cause confusion among workers familiar with the literature on the Middle American segment of the E. goudotii complex in which the "old" head scale terminology is solidly entrenched and needed no change for that group.
Morphological variation among all leptotyphlopid snakes is summarized in Adalsteinsson et al. (2009). Abbreviations used here are: SVL = snout-vent length; TAL = tail length; and TOL = total length. TOL was measured by laying the snake alongside a standard 12 inch ruler containing mm (millimeters). All remaining measurements were made using dial calipers measured to the nearest 0.1 mm. Color statements given are those in alcohol unless otherwise noted. Because we are not aware of any external morphological characters to distinguish males from females, both sexes were combined for all morphological characters as was done for E. goudotii by Pinto et al. (2010). Also, Hedges (2008) found that New World leptotyphlopids are less sexually dimorphic than Old World species. The synonymies presented herein include the first use of any combination that pertains to the species in question, including all known misspellings, and the first reference to the single i and double ii spellings in various name combinations. Type numbers given for new species proposals and their stated type localities are also given for each new species proposal. Museum acronyms follow Sabaj Pérez (2014) and color names (capitalized) and numbered color codes (in parentheses) are those of Smithe (1975Smithe ( -1981. In the following list of specimens examined, one asterisk denotes specimens from which external morphological data were taken whereas two asterisks designate specimens that were recorded only for the frontal scale condition. GenBank accession numbers are listed from (Adalsteinsson et al. (2009); in the order 12SrRNA, cytochrome b; n/a = gene not sequenced) for the old samples as well as the new genetic samples generated for this study.

RESULTS
The molecular phylogeny (Fig. 2), which includes new sequence data and localities, supports the recognition of four taxa closely related to E. goudotii that have previously been described (Cope, 1875;Oliver, 1937;Taylor, 1940). Together with E. goudotii, these five species are supported as well by external morphological characters as described in the sections below (molecular data not available for Colombian E. goudotii, the country in which the type locality of E. goudotii occurs). As pointed out above, these five species were also recognized by Wallach, Williams & Boundy (2014), but those authors did not provide any evidence or comments to support their decisions.
The phylogenetic tree (Fig. 2) shows the same general pattern found by Adalsteinsson et al. (2009), namely, that Epictia albifrons is only distantly related to other species of the genus, the latter of which form two distinct clades. The Caribbean clade contains a pair of species (E. columbi and E. magnamaculata), whereas the Middle American Clade contains three species recognized here (E. ater, E. bakewelli, and E. phenops). One species in the Caribbean clade, E. columbi, is known only from a small Bahamian island (San Salvador) and thus it is not surprising that there is no genetic differentiation among those five individuals. However, it was unexpected to find, also, that there was little genetic difference among samples of E. magnamaculata, with localities separated by 280 km of open sea (Swan Islands and Bay Islands). The genetic structure within the Middle American clade is more pronounced (Fig. 2), showing deep divergences within E. bakewelli and E. phenops. Although these groups are also diagnosable morphologically, and are cohesive geographically (see below), the genetic differences within those two species are as great as that between E. columbi and E. magnamaculata, suggesting that there are additional species of Epictia in Middle America not yet recognized.
Based on the genetic results and the presence of a relatively broad dark dorsolateral stripe (see Peters & Orejas-Miranda, 1970)  apparently completely allopatric with E. goudotii, the latter, in our opinion (but too few genetic data and museum specimens available to us to do good analyses), occurring from coastal Venezuela, across northern Colombia, to the region of the Panamanian Canal Zone, and on Trinidad (see Materials and Methods). The remaining three mainland species treated herein (E. ater, E. bakewelli, and E. phenops) have a collective distribution from northwestern Costa Rica into Mexico as far north as coastal Jalisco on the Pacific versant and southern Tamaulipas, Mexico, on the Atlantic versant. Brief reviews of the synonymy, geographical distribution, diagnosis, variation, and remarks are provided for each of the five species of the E. goudotii complex studied herein. Neither E. bakewelli nor E. phenops have had good morphological descriptions published, but since all five species of the E. goudotii complex covered in this work are similar in overall external morphology, we only provide a detailed morphological redescription of E. phenops.
Leptotyphlops phenops: Smith (1939:28).  Diagnosis. Epictia phenops is one of the three species of the E. goudotii complex under study herein that lack rostral-frontal fusion (= has a frontal scale present). Epictia ater and E. bakewelli are distinguished from E. phenops in having rostral-frontal fusion. Epictia goudotii and E. magnamaculata are the two species that agree with E. phenops in having a frontal scale (= lack rostral-frontal fusion). Epictia phenops differs from E. goudotii in having the pale tail spot almost always larger ventrally than dorsally, covering 0-1 scales dorsally and 7-15 scales ventrally (versus pale tail spot, when present, larger dorsally than ventrally, covering 1-4 scales dorsally, 0-1 scales ventrally in E. goudotii). Epictia phenops differs from E. magnamaculata in usually having indistinct dark brown body stripes on a paler brown ground color, an indistinct to absent pale snout spot that is usually confined to the rostral when present, and a usually distinct pale tail spot that is almost always larger ventrally than dorsally (versus distinct alternating black and dark brown zig-zag body stripes, pale snout spot distinct with the spot almost always extending from the rostral onto adjacent edges of upper nasal scales, and a distinct pale tail spot that is almost always larger dorsally than ventrally, covering 2-6 scales dorsally and 2-5 scales ventrally in E. magnamaculata). The rare specimens (3 of 54 individuals) of E. phenops that have rostral-frontal fusion can be difficult to distinguish from E. ater and E. bakewelli, except that some E. phenops tend to have the anterior third of the venter paler brown than the adjacent dorsum (these surfaces usually about same color in E. ater).
Remarks. Cope (1875:128) stated that the new species Stenostoma phenops "is represented by numerous specimens" with most having come from "Tehuantepec" and at least one was from "Coban, Guatemala." Thus, the series on which Cope designated his new species became syntypes. Cochran (1961:214) listed nine USNM "Cotypes," eight from Tehuantepec and one from Cobán, Guatemala. All nine specimens listed by Cochran still exist today in the USNM collection. Among those specimens, we designate USNM 30290 as the lectotype. The single measurement given by Cope most closely resembles USNM 6760 from Cobán, Guatemala. However, based on a review of the literature, we feel it is best to designate a specimen from Tehuantepec as the lectotype where eight of the nine "cotypes" listed by Cochran (1961) are from. USNM 30290 is an adult in good condition, with a TOL of 147 mm, lacks rostral-frontal fusion, has 244 median dorsal scales, 13 medial subcaudal scales, the pale snout spot confined to the rostral, the pale tail spot much larger ventrally than dorsally, thus agrees well with Cope's description of the species. Thus, the remaining "cotypes" listed by Cochran (1961) become paralectotypes (USNM 6760, 12444, 30289, 30291-95) of Stenostoma phenops Cope. The type locality restrictions by Smith & Taylor (1945:24) & Smith & Taylor (1950 to "Tehuantepec" and "Tehuantepec (city, and environs)," respectively, were made without supporting data, thus are invalid. Mertens (1952) wrote that one of the five specimens of Epictia phenops he examined from nearby localities in El Salvador had the rostral fused with the frontal scale (as praefrontale). Reexamination of those five specimens shows that one (SMF 43218) now lacks a head and the remaining four all have a frontal scale, although the suture between the rostral and frontal is indistinct in one (SMF 43216). Our specimens examined include all of the USNM specimens of Epictia phenops included by (Smith (1943), except for USNM 30094 (now soft and in bad condition), 110312 (exchanged), and 110318 (lost)), but with new scale counts in an effort to have a single person making those counts.
Pinto et al.  Oliver (1937; as Leptotyphlops albifrons from Tehuantepec (=Epictia phenops)) included a drawing of the dorsal head scales of this species that shows the presence of the diagnostic frontal (= prefrontal of old terminology) scale, which should be helpful to the reader in visualizing that important character, especially considering the new head scale definition that will likely cause confusion since all previous workers on Middle American E. goudotii complex members have consistently used the old head scale terminology.
Leptotyphlops albifrons: Nicéforo Maria (1942:86). Geographic distribution. Epictia goudotii is known to occur at low and moderate elevations from the Canal Zone, Panama, across northern Colombia and Venezuela, and on Trinidad (but see Remarks). See Materials and Methods for a list of specimens examined and their locality data (also see Fig. 4).
Diagnosis. Epictia goudotii, along with E. magnamaculata and E. phenops, are the three species of the E. goudotii complex under study herein that have a frontal (prefrontal) scale. Epictia goudotii differs from E. magnamaculata in having the pale snout spot very indistinct to distinct, with that spot usually confined to the rostral scale when distinct, and the pale tail spot nearly absent to indistinct (versus distinct pale snout spot almost always extending onto the adjacent edges of the upper nasal scales and the pale tail spot always distinct in E. magnamaculata). Epictia goudotii differs from E. phenops in having the pale tail spot nearly absent to indistinct, that spot almost always larger dorsally than ventrally, covering 1-4 scales dorsally and 0-1 scales ventrally (versus pale tail spot usually distinct and larger ventrally than dorsally, covering 0-1 scales dorsally and 7-15 scales ventrally in E. phenops). Epictia goudotii differs further from E. ater and E. bakewelli, the two species that lack a frontal (prefrontal) scale (= have rostral-frontal fusion) as follows: from E. ater in having the pale tail spot almost always larger dorsally than ventrally (versus tail spot, when present, much larger ventrally than dorsally in E. ater); from E. bakewelli in having the ventral surfaces about the same color as the dorsal surfaces (versus ventral surface of head and anterior third of venter distinctly paler than adjacent dorsum in E. bakewelli).
Leptotyphlops goudotti: Wilson & Meyer (1985:20) (in part). Geographic distribution. Epictia magnamaculata occurs at low elevations on the following Caribbean islands: the Bay Islands, including the Cayos Cochinos, Honduras; Islas del Cisne (Swan Islands), Honduras; Cozumel, Mexico; and San Andrés and Providencia, Colombia. See the Materials and Methods for a list of specimens examined and their locality data (also see Fig. 4).
Diagnosis. Epictia magnamaculata, along with E. goudotii and E. phenops, are the three species of the E. goudotii Species complex studied herein that have a frontal (prefrontal) scale. Epictia magnamaculata differs from E. goudotii in always having distinct black and brown body stripes on an overall black ground color, a distinct pale snout spot that almost always extends onto the adjacent edges of the upper nasal scales, and a distinct pale tail spot (versus pale snout spot very indistinct to distinct, spot usually confined to rostral, extending onto adjacent edges of upper nasal in one of five; pale tail spot nearly absent to indistinct in E. goudotii). Epictia magnamaculata differs from E. phenops in always having distinct black and brown body stripes on an overall black ground color, a distinct pale snout spot that almost always extends onto the adjacent edges of the upper nasal scales, and a distinct pale tail spot that is usually larger dorsally than ventrally (versus dark brown body stripes usually indistinct, pale snout spot very indistinct to distinct with the spot usually confined to rostral when distinct, and pale tail spot larger ventrally than dorsally in E. phenops). The rare E. magnamaculata specimen that lacks a frontal scale (3 of 58 individuals) differs from the two species that also lack a frontal (prefrontal) scale (E. ater and E. phenops) in having distinct black and brown body stripes, a distinct pale snout spot that almost always extends onto the adjacent edges of the upper nasal scales, and a distinct pale tail spot.
Variation. Pinto et al. (2010) gave the following morphological ranges for the specimens they examined from Islas de Providencia and San Andrés, Colombia: middorsal scales 245-262 (n = 12); midventral scales 237-246 (n = 4); subcaudal scales 15-18 (n = 13); TL 98-195 (n = 7) mm; Remarks. The type description of Leptotyphlops magnamaculata provided by Taylor (1940) is fairly detailed and accurate, but based on only the holotype. Pinto et al. (2010) also provided information on morphological variation in a series (ca. 13) from the Colombian islands of Providencia and San Andrés and McCranie (2011) gave a detailed morphological description of this species (using the traditional head scale terminology of prefrontal as used in the Middle American E. goudotii complex) based on a series of 31 specimens from Honduran islands. Taylor (1940) provided a head drawing of the holotype of Epictia magnamaculata showing the presence of the frontal (prefrontal) scale and the typical pale snout spot size. As evidenced from the non-overlap of some scale counts of E. magnamaculata from Islas de Providencia and San Andrés, Colombia versus specimens from the Bay Islands of Honduras, the former likely represents an undescribed species that we are currently studying. Wallach, Williams & Boundy (2014:277) proposed Leptotyphlops albifrons margaritae Roze to be a synonym of Epictia magnamaculata, but without offering any supporting data. The description and illustrations of the head scales, along with the color description, of that nominal form provided by Roze (1952:154-156), however, are not convincing enough for us to agree that E. a. margaritae is a junior synonym of E. magnamaculata.
Leptotyphlops goudotti phenops: Villa (1983:37). Geographic distribution. Epictia ater occurs at low and moderate elevations from western Honduras to northwestern Costa Rica, including Islas Murciélago (see Remarks). See the Materials and Methods for a list of specimens examined and their locality data (also see Fig. 4).
Diagnosis. Epictia ater, along with E. bakewelli, are the two species of this complex under study herein to have the rostral fused with the frontal (prefrontal) scale, thus the fused rostral-frontal scale contacts the postfrontal scale. Epictia ater differs most obviously from E. bakewelli in having the ventral surfaces essentially the same color as the dorsal surfaces (versus underside of head and about anterior third of venter distinctly paler brown than the brown dorsal surfaces in E. bakewelli). Epictia ater also differs from E. bakewelli in having the dark body stripes absent or indistinct (versus those stripes distinct in E. bakewelli). The single E. ater specimen with a frontal scale differs from two of the three species that also have a frontal scale, E. goudotii and E. magnamaculata, as follows: from E. goudotii in having the pale tail spot, when present, much larger ventrally than dorsally (versus tail spot larger dorsally than ventrally when present in E. goudotii); from E. magnamaculata in lacking distinct black body stripes and having the pale tail spot, when present, much larger ventrally than dorsally (versus distinct black stripes present and pale tail spot usually larger dorsally than ventrally in E. magnamaculata). A specimen of E. ater with a frontal scale can be difficult to distinguish morphologically from the normal E. phenops.
Remarks. The type descriptions of Epictia ater and its synonym E. nasalis provided by Taylor (1940) are fairly detailed and accurate. Unfortunately, external morphological variation in the species was poorly documented until McCranie (2011) provided a detailed description of E. ater based on Honduran specimens. However, the McCranie (2011) description also included data from four specimens (of 28 included) now known to represent E. phenops.
No museum specimens of Epictia ater from Costa Rica were available to us at the time we were borrowing museum specimens. However, Savage (2002:559) stated the E. (as Leptotyphlops) ater from Costa Rica were in agreement with the E. ater of Nicaragua (including the holotype) in having "rostral-prefrontal fusion" (= rostral-frontal fusion) and were also in agreement with the color pattern of the Nicaraguan specimens. For those reasons, Savage (2002) elevated E. ater to a valid species for the Nicaragua and Costa Rica populations. As noted in McCranie (2011) and in this work, most Honduran populations agree with those characters and are assigned with the Nicaragua and Costa Rica populations to the species E. ater. Thus, without our examination of Costa Rica specimens, we are confident in assigning Costa Rican specimens to E. ater along with those of Nicaragua and Honduras. The range of E. ater terminates in northwestern and north-central Costa Rica. Savage (2002) gave USNM 79947 for both holotypes of Epictia ater and its synonym E. nasalis, probably because Taylor gave the correct USNM number (79947) in the legend for his Fig. 4, but gave an incorrect number in his list of the designation of the holotype. Taylor (1940) included head drawings of E. ater, and its synonym E. nasalis, showing the absence of the frontal scale. Wallach, Williams & Boundy (2014) included the El Salvadoran and western Honduran populations as E. ater (in error). However, the molecular and external morphological data for those populations from El Salvador and Copán, Honduras, suggest they are E. phenops. Because of the overall similarity in dorsal color pattern of E. ater with that of E. phenops, no photograph of E. ater is included herein.
Geographic distribution. Epictia bakewelli occurs at low and moderate elevations from Colima to the foothills west of the Isthmus of Tehuantepec, Oaxaca, Mexico. See the Materials and Methods for a list of specimens examined and their locality data (also see Fig. 4).
Diagnosis. Epictia bakewelli, along with E. ater, are the two species of this complex to have the rostral fused with the frontal (prefrontal) scale, thus the fused rostral-frontal scale contacts the postfrontal scale. Epictia bakewelli differs most obviously from E. ater in having the underside of head and about the anterior third of the venter distinctly paler brown than the adjacent brown dorsal surfaces (versus ventral surfaces essentially the same color as the dorsal surfaces in E. ater). Epictia bakewelli also differs from E. ater in having distinct dark body stripes (versus those stripes indistinct or absent in E. ater). Epictia bakewelli differs further from two of the three species that have a frontal scale (E. goudotii and E. magnamaculata) in having the underside of head and about anterior third of the venter distinctly paler brown than the adjacent brown dorsal surfaces and the pale tail spot larger ventrally than dorsally (versus those ventral surfaces similar to dorsal surfaces and the pale tail spot larger dorsally than ventrally in E. goudotii and E. magnamaculata). The rare specimen of E. phenops that lacks a frontal scale (3 of 54) can be difficult to distinguish from E. bakewelli.
Variation. Our data from Epictia bakewelli are based on only ten specimens examined (see Materials and Methods) plus limited data extracted from the Oliver (1937) & Duellman (1956 descriptions of the holotypes of E. bakewelli and its synonym E. gadowi, respectively: middorsal scales 226-262 (242.1 ± 10.7, n = 12); midventral scales 206-244 (223.0 ± 11.1, n = 11); subcaudal scales 14-20 (17.9 ± 2.1, n = 11); TL n = 11) mm;, n = 11) mm; TL 4.7-8.0 (n = 11) % of TOL; frontal absent in all 12; supraocular extending anteriorly to about level of mideye to just anterior to mid eye; supraocular not contacting anterior supralabial; anterior supralabial extending dorsally only to level of lower third of eye; pale snout spot absent to distinct, usually confined to rostral when present (extending onto adjacent edges of upper nasal in two of ten); pale tail spot distinct, larger ventrally than dorsally, covering 0-4 scales dorsally (confined to tail spine in two of ten), and 1-11 scales ventrally; dark stripes on body distinct in all, zig-zag shaped; under surface of head and at least adjacent third of belly much paler brown than adjacent dorsum.
Remarks. The type descriptions of Epictia bakewelli and its synonym E. gadowi (both as Leptotyphlops) were brief and lacked much standard scale data and an adequate study of the variation in scales in E. bakewelli has not been published. Oliver (1937) & Duellman (1956 provided dorsal head drawings of the holotypes of Epictia bakewelli and E. gadowi (a synonym of E. bakewelli), respectively. Those drawings show the absence of the frontal scale, one of the diagnostic characters of this species. Wilson & Hahn (1973) suggested placing E. bakewelli in the synonymy of E. phenops based on variation in the presence or absence of rostral-frontal fusion (= presence or absence of a frontal scale), but that action was based on little evidence and also ignored the detailed color pattern of E. bakewelli given by Oliver (1937). The color pattern, along with the absence of the frontal scale are the two most important morphological diagnostic characters to distinguish E. bakewelli from the remaining members of the E. goudotii Species complex. No quality photographs of E. bakewelli were available to us to include herein.
One of the Epictia bakewelli specimens examined by Smith (1943;USNM 110306) has been subsequently lost and another (USNM 30295) identified as Leptotyphlops phenops bakewelli by Smith (1943) has been reidentified as E. phenops. Recently, Wallach, Williams & Boundy (2014:276) recognized E. bakewelli as a species, but provided no data to support that decision.

DISCUSSION
Here, we revise the Epictia goudotii Species complex to include six species: E. ater, E. bakewelli, E. columbi (not studied morphologically herein), E. goudotii, E. magnamaculata, and E. phenops. Adalsteinsson et al. (2009) showed with molecular data that E. albifrons is not closely related to these species and is therefore not a member of the Epictia goudotii Species complex, even though it has been confused with E. goudotii in the past. Therefore, Epictia albifrons/E. tenella and relatives should be considered a separate complex, the Epictia albifrons Species complex. Although we lacked genetic data for E. goudotii, we consider it to occur from the Panama Canal Zone to northern Colombia and Venezuela (see also Pinto et al., 2010).
Molecular clock estimates of divergence times (Adalsteinsson et al., 2009) indicate that the Epictia goudotii Species complex split from E. albifrons in the early Cenozoic and that the Caribbean and Middle American Clades of the complex diverged from one another in the Oligocene (∼30 million years ago). Even closely related species, such as E. columbi and E. magnamaculata, appear to have diverged more than 10 million years ago (Adalsteinsson et al., 2009). These deep splits provide further support for the recognition of species, rather than subspecies, in this Species complex, and suggest that additional species remain to be recognized given the phylogenetic structure within E. bakewelli and E. phenops. The presence of E. columbi in the Bahamas, in itself, indicates that these snakes can disperse over ocean waters. Additional support for oceanic dispersal comes from this deep timeline (Adalsteinsson et al., 2009), which infers that Middle America was invaded by these snakes (from South America) prior to the emergence of the Isthmus of Panama approximately 3 million years ago.
As noted in the introduction, the taxonomic history of the Epictia goudotii complex has been complicated and confused. Epictia albifrons and E. tenella lie outside the scope of this work, and because they are not closely related to the E. goudotii complex (Adalsteinsson et al., 2009), are discussed herein only because we used sequences tentatively identified as E. albifrons/E. tenella in our molecular phylogeny, as an outgroup. Epictia albifrons has also been confused with E. goudotii in some earlier literature. Some debate occurs in the literature about the validity of one or both of those nominal forms. Both have sometimes been recognized as valid species (Wallach, Williams & Boundy, 2014; but without documentation), or E. albifrons was considered a nomen dubium (Franco & Pinto, 2010), or E. tenella was considered a junior synonym of E. albifrons (Hoogmoed & Gruber, 1983). Klauber (1939:59) diagnosed E. tenella from E. albifrons by the "contact between supraoculars and the anterior supralabial" in E. tenella versus the absence of that contact in E. albifrons because "the junction of the nasal and ocular" prevented such contact. The destroyed holotype of E. albifrons (but see neotype designation in Mumaw, González & Fernández, 2015) along with the poor description of the holotype and the possibly erroneous type locality data all contribute to this problem.
Orejas-Miranda (1967) considered both Epictia albifrons and E. tenella to have large geographical distributions that included Guyana. Orejas-Miranda (1967) noted the absence versus presence of the supraocular-anterior supralabial contact in those two species, but could not find any museum specimens representing, in his opinion, E. albifrons. Peters & Orejas-Miranda (1970) recognized both E. albifrons and E. tenella as valid species, and also noted the presence (E. tenella) or absence (E. albifrons) of the supraocular-anterior supralabial contact, but included only E. tenella in the fauna of the "Guianas." Hoogmoed (1977) studied the threadsnakes from Suriname and assigned the name Epictia (as Leptotyphlops) tenella to those populations. Hoogmoed & Gruber (1983) suggested placing E. tenella in the synonymy of E. albifrons in order to provide a much-needed diagnosis of E. albifrons. Hoogmoed & Gruber claimed that a combination of characters they could see in Wagler's (1824) illustration of Stenostoma albifrons strengthened their opinion, but we think that Wagler's illustration is too poor to be of any use in species identification and the Hoogmoed & Gruber (1983) decision to place E. tenella as a junior synonym of E. albifrons is premature. Klauber (1939:61), regarding the supraocular-anterior supralabial contact, stated, "The available specimens indicate that in this area (Guyana and Trinidad) the character is quite consistent, although in one specimen (Carnegie Museum 4890) these scales fail to make contact on the right side of the head." Klauber (1939:61) also said, "Among the specimens of the albifrons-group which I have had available from South and Central America these from British Guiana and Trinidad have been the only ones having the anterior supralabials in contact with the supraoculars." Klauber (1939) recorded E. tenella from Trinidad (with AL-SO contact), but the single Trinidad specimen examined for the present work (USNM 12498) lacks that contact. Considering that we found almost no variation, and what occurred was minor, in the sizes and extent of the anterior supralabial and supraocular in the large series of Epictia we examined (except the aberrant absence of the supraocular in the holotype of E. nasalis; but see Dunn & Saxe (1950) who noted variation in this character in 2 of 5 specimens from Nicaragua that were not examined by us), we doubt that significant variation occurs in which the supraocular can regularly extend anteriorly to contact the anterior supralabial in one specimen but not in another of the same species. We agree more with the statement by Thomas (1965:6) that "There is obviously more than one species involved in the material (of E. albifrons) I have seen" when he was discussing "South American mainland" specimens he examined. Recently, Mumaw, González & Fernández (2015) considered E. tenella to represent a Species complex. Clearly, a combined DNA and morphological study of South American populations of this complex is needed, especially one that uses both types of data sets with northern South American populations that contain both the anterior supralabial-supraocular contact and those that lack that contact. However, current export prohibition of tissues from several South American countries is a serious obstacle to such a study. Also, it is unfortunate that Pinto et al. (2010) did not study genetic data in their review of Colombian Epictini as Colombian tissues are not available for export. Franco & Pinto (2010) also did not include genetic data in their work on the E. albifrons/E. tenella problem.