Higher-Level Taxonomy

The vast majority of Neotropical colubrid snakes have long been considered to belong to the poorly characterized subfamily Xenodontinae. Two large assemblages have been informally called Central American and South American xenodontines, although there is broad geographic superposition. These two groups received suggestive support as clades from early microcomplement fixation studies of serum albumins (Cadle, 1984a, 1984b, 1984c, 1985). Some hemipenial differences were pointed out by Myers and Cadle (1994: 27–28), who thought that the hemipenes of the South American clade were relatively primitive compared with the Central American clade, which is characterized by several derived hemipenial features.

Zaher (1999) recognized the mainly South American clade as Xenodontinae, sensu stricto, based on two hemipenial synapomorphies (enlarged calyces on the asulcate sides of the lobes, and enlarged lateral spines), which are seen either individually or together in most species. Zaher (1999) recognized the mainly Central American clade as subfamily Dipsadinae, based on several hemipenial synapomorphies suggested by Myers and Cadle (1994: 27).

Zaher et al. (2009) subsequently reworked the classification of caenophidian snakes based on a new molecular phylogeny. The family Colubridae was greatly restricted from “its long-standing use [for] all caenophidians that were not acrochordids, elapids, or viperids” (Zaher et al., 2009: 132). In addition to confirming removal of a number of Old World groups whose relationships had been questioned, the name Colubridae in its common sense was most significantly affected by raising its main subfamilies (colubrines, natricines, dipsadines) to familial level. The subfamily Xenodoninae was returned to its earlier concept of a single group of New World “xenodontines” by submersing it within family Dipsadidae. The Xenodoninae were then judged to have no known synapomorphies because those presented by Zaher (1999) were moved to the Dipsadidae (see Zaher et al., 2009: 140). The Xenodoninae were nonetheless retained as a subfamily in order to avoid “changing the well-established taxonomic hierarchy for this group” (Zaher et al., 2009: 141–142).

Myers (2011: 11ff.) responded narrowly to Zaher et al. (2009) by arguing that the aforesaid familial changes were unnecessary. He believed that “discussion is hampered and becomes confused when new taxonomies are generated from new (uncorroborated) phylogenetic hypotheses, especially when familiar groups are renamed and redefined in major ways” (Myers, 2011: 12). As a practical reason for retaining Colubridae, sensu lato, Myers (2011: 13) noted that he could not unambiguously assign the new genus Amnesteophis to any of four subfamilies to which he had compared it.

We agree with Zaher and others that sound classification should reflect phylogeny, but we disagree fundamentally as to whether taxonomy needs to be changed almost automatically with new phylogenetic insight. We value any insights provided by the molecular phylogeny generated by Zaher et al. (2009), but for practical reasons (sensu Myers, loc. cit.) we continue to recognize the Dipsadinae, Xenodontinae, and Colubrinae as subfamilies of Colubridae. We continue to use “colubrid” as a sort of common name for snakes in or closely related to members of these subfamilies. We follow Zaher (1999: 3) in using, for pragmatic reasons, the term “xenodontines” in quotes for many dipsadine and xenodontine genera of uncertain assignment, including some colubrids used herein for anatomical comparisons (e.g., Diadophis, Rhadinophanes, Taeniophallus).

It must not be thought that our few critical comments are directed solely at Zaher et al. (2009); we are suspicious of all phylogenies that are published rapidly in order to yield “provisional” taxonomies (e.g., Kelly et al., 2011; see fn. 17 herein). Although new genera and new species must be dealt with in a timely fashion, “provisional” higher-level classifications can and should compete among themselves before new taxonomies are accepted by biologists who actually use them. Nonsystematists need at least “provisional” stability; we find it admirable when authors of molecular phylogenies (e.g., Vidal et al., 2008) decline to make taxonomic decisions straightaway whenever data are insufficient for resolving lineages adequately. Snake taxonomy remains unstable and more than ever it is in a state of change, and so should be approached as flexibly and as conservatively as possible.

The new taxa are named and described forthwith. Additional global comparisons relevant to the taxonomy and intensive discussion, including resurrection of a few previously named cryptic species, are given in sections following.

Description of Type Specimens

Eutrachelophis bassleri is a small slender snake that probably is sexually mature by 300 mm total length, with a maximum known length of 377 mm. Following is a combined description of the 12 specimens in the type series.

Hemipenis

Remarks

The general location of Bassler's “Pisqui Hills”—the type locality of Eutrachelophis bassleri—can be determined from an unpublished report (Bassler, ms.). Bassler first explored and mapped the Río Pisqui in 1923, traveling 186 km by river from its mouth (149 m elev.) to the head of canoe navigation (259 m), followed by an additional 18 km on foot beyond the first cataracts—a straight-line distance of 93 km by Bassler's careful reckoning. He determined latitude and longitude as 7°42′ S, 75°00′ W at the mouth of the Río Pisqui and as 8°22′ S, 75°30′ W at the end of his traverse; it may be noted that the first set of coordinates agrees within a few minutes to recent maps and gazetteers. Bassler described the first 45 km (by river) as “a flood plain subject for the greater part to a period of inundation each year and the channel here is not permanent for lateral erosion is active.” Upriver, “this recent flood plain merges into a plain determined by firmer though still practically unconsolidated sediments and here the channel is deeper and appears to change very little even over long periods of time.” Somewhere on this upper plain Bassler wrote that,

The 200 ft-high (60 m) hills on the higher flood plain of the upper Río Pisqui must certainly be what Bassler gave as “Pisqui Hills” for specimens collected on this and subsequent trips, and the locality can with confidence be placed within the area bounded by parallels 8°00′–8°22′ S and meridians 75°30′–75°50′ W.

Fig. 4.

Eutrachelophis species (MZUSP 10530, head in dorsal and lateral view; a female from Cabeceira do Rio Urucu, central Amazonian Brazil). This unnamed snake appears to be the sister species of E. bassleri. See Remarks and compare with figure 2. Formalin preservation caused the integument of this specimen to become slightly translucent, giving visibility to a large “supralabial” gland (SG); this large gland (similarly positioned in all three species of Eutrachelophis) is adherent to the medial side of the supralabial integument. The mucous and serous (Duvernoy's) parts of the gland cannot be distinguished in Eutrachelophis without histological examination.

In 1993, the late Paulo E. Vanzolini sent to Myers what he recognized as a new species of snake from Cabeceira do Rio Urucu, Amazonas, Brazil. It clearly is closely related to Eutrachelophis bassleri and seemed likely to be a different species. But it could not be so identified with assurance because (1) ocellar head and neck patterns in Eutrachelophis and other genera are somewhat variable, and (2) because it is a female collected at about 5°S, 65°W—far to the east of known bassleri localities (see map 1). Ana Lucia C. Prudente fortunately has obtained additional material that she will describe separately.

Fig. 5.

Eutrachelophis steinbachi (Boulenger). Adult female shown approximately life size (BMNH 1946.1.21.62, lectotype by present designation).

Fig. 6.

Eutrachelophis steinbachi (Boulenger). The paralectotype (BMNH 1946.1.21.63), a juvenile male shown 1.6× life size.

Fig. 7.

Eutrachelophis steinbachi (Boulenger), handheld preserved specimens. Upper. A specimen (NMW 23106) cited as “syntypus” by Tiedemann and Häupl (1980: 61) but not mentioned in Boulenger's (1905: 454) original description. Lower. The syntype (BMNH 1946.1.21.63) designated lectotype in this paper (photograph courtesy of James R. Dixon).

Fig. 8.

Head of Eutrachelophis steinbachi (Boulenger) (UMMZ 60736), showing details of color pattern. The oblique alignment of the pale ocelli differentiates E. steinbachi from E. bassleri and its unnamed sister species (cf. figs. 2 and 4).

Fig. 9.

Uneverted hemipenis of Eutrachelophis steinbachi (Boulenger); right organ of AMNH R-125695, opened by midventral incision, ×5. Abbreviation: STB, smooth terminal basin.

Fig. 10.

Skull of Eutrachelophis bassleri, new species (AMNH R-55786), 38.8. Abbreviations: A, angular; B, basioccipital; CB, compound bone; ct, foramen for chorda tympani of facial nerve; ic, foramen for internal carotid artery; D, dentary; E, ectopterygoid; EX, exoccipital (+ opisthotic); F, frontal; ic, foramen for internal carotid artery; M, maxilla; N, nasal; OF, orbital fenestra, containing ophthalmic (V₁), oculomotor (III), trochlear (iv) nerves; PA, parietal; PF, prefrontal; PL, palatine; PM, premaxilla; PO, postorbital; PR, Prootic; PT, pterygoid; pv, foramen for pituitary vein and retractor pterygoideus nerve; Q, quadrate; S, stapes; SM, septomaxillary; SO, supraoccipital; SP, sphenoid (fused basisphenoid and parasphenoid); SPL, splenial; T, tabular; V, vomer; Vc, foramina for Vidian canal; Vlp, foramen for levator pterygoideus ramus of trigeminal nerve; Vpt, foramen for protractor pterygoideus ramus of trigeminal nerve; Vrp, foramen for retractor pterygoideus ramus of trigeminal nerve; V2, foramen for maxillary ramus of trigeminal nerve; V2VIIp, foramina for combined maxillary ramus of trigeminal nerve and palatine ramus of facial nerve ( =  orbitopalatine nerve); V3VII, foramen for mandibular ramus of trigeminal nerve and facial nerve; VIIp, foramina for palatine ramus of facial nerve; IX, foramen for glossopharyngeal nerve; X, foramen for vagus nerve; XII, foramina for hypoglossal nerve.

Fig. 11.

Skull of Eutrachelophis steinbachi (Boulenger) (AMNH R-125695), ×7.7. Abbreviations: Same as for figure 10, but here repeated only for nerve and venous foramina. With only one skull each of E. bassleri and E. steinbachi, no significance can be attached to any slight differences in positioning of these foramina, which are so labile that even the left and right sides of the same skull may sometimes differ.

Fig. 12.

Maxillary dentition of Eutrachelophis steinbachi (Boulenger). Right maxilla of AMNH R-125695, in lateral and ventral view. Arrangement of tooth sockets shows a slight lateral offset of the last tooth behind the diastema; the offset is more pronounced in most other “xenodontines,” although a few genera have the posterior maxillary teeth arranged in a straight line configuration.

Rhadinaea Steinbachi Boulenger, 1905: 454 (two syntypes, female and young, from “the Province Sara, Department Santa Cruz de la Sierra, collected by Hr. J. Steinbach”).

Aporophis melanocephalus Griffin, 1916: 171–172 (holotype CM R-18, “a female, taken at Las Juntas, Bolivia, 250 M. above sea-level, by José Steinbach in December, 1913”).

Rhadinea steinbachi Boulenger: Dunn, 1922: 220 (Aporophis melanocephalus placed in synonymy).

Liophis steinbachi (Boulenger): Amaral, “1929b” [1930]: 174. Peters and Orejas-Miranda, 1970: 179. Myers, 1974: 22 (comment on unsatisfactory generic assignment); Myers and Cadle, 1994: 2. Dixon, 1980: 15, 20 (listed as incertae sedis).

Rhadinea steinbocki (misspelling): Clark, 1945: 428 (mention of hemipenis).

Description

Hemipenis

The following description is entirely based on retracted hemipenes (fig. 9). The uneverted left organs of AMNH R-125695 and FMNH 35662 had been opened midventrally; they were removed and pinned flat for detailed study and illustration. Supplementary notes were provided by examination of in situ organs in UMMZ 60734 and 63216.

Major retractor muscle originating at level of subcaudal 37 for the right hemipenis of FMNH 35662, anteriorly dividing at the levels of subcaudals 20 (1 specimen), 19 (2), or 15 (1), and inserting on the ends of the hemipenial lobes at subcaudals 17 (1), 15 (2), or 14 (1). The hemipenis therefore is relatively long, spanning 14–17 subcaudals. The two lobes are narrow and long, comprising nearly two-thirds the total length of the hemipenis, which bifurcates at the level of subcaudal 6 in three specimens and at subcaudal 7 in another.

The extreme base of the organ is virtually nude except for a sparse distribution of spinules; a relatively deep basal groove on the dorsal wall might persist on the everted hemipenis as a basal naked pocket (sensu Myers, 1974: 32), but this is uncertain. Two large and two medium-sized spines arise across the middle of the undivided base of the organ; these spines are nearly straight, and the tips of the two largest extend nearly to the base of the hemipenis, on either side of the sulcus (fig. 9). Above the big spines are numerous small, straight to slightly recurved spines, those close to the sulcus being very small. The proximal 40%–50% of each hemipenial lobe is profusely covered by short, thick spinules. The distal 50%–60% of a lobe is densely covered with straight, relatively long thin spines, and the sulcus branch ends abruptly in this region, at a length about 80% up the lobe.

There is at the base of the lobes an interlobular nude space—the STB, or “smooth terminal basin” (see Hemipenial Terminology for Snakes)—which is closely edged by small, short spines (fig. 9). This “basin” has apparent expansion folds and therefore lacks the sulcuslike smoothness seen in some other taxa (e.g., compare with the STB shown in fig. 18). The expansion folds suggest that the basin will be considerably enlarged after eversion.7

Outside the STB, the long hemipenial lobes are completely spiny as described above. The sulcus proximally lies on the lateral wall (both on left and right organs) and divides halfway up the base, at the level of subcaudal 3. The branches of the sulcus are deeply incised and have a centrifugal orientation, lying on the ventral wall of the ventral lobe and on the dorsal wall of the dorsal lobe; each sulcus branch terminates well short of the apex.

Remarks

As indicated above, Eutrachelophis steinbachi (Boulenger) is known to us from a dozen specimens. Excluding one specimen of questionable provenance, all originated over a period of time (circa 1903–1928) from the Bolivian collectors José Steinbach and his son Francisco Steinbach. Their locality designations are of a somewhat general nature, but the several Steinbach localities for this species seem to lie at the eastern base of the Andes in the department of Santa Cruz.

Buenavista ( =  Buena Vista of modern maps and gazetteers) and Provincia Sara are the only localities for Eutrachelophis steinbachi shown on map 1, although the old “Province Sara” [Provincia Gutiérrez]—the type locality—is a region rather than a single collecting site.8 José Steinbach lived in Buena Vista (17°27′S, 63°40′W), the most frequently mentioned locality. His locality Río “Sirutu” for a UMMZ specimen may be one of several spelling variants of Río Surutú, a known Steinbach locality (Paynter, 1992: 142–143). According to Sydney Anderson (personal commun.), “The most probable collecting areas near the Río Surutú would have been reasonably near the town of Buenavista … probably to the west or southwest of that village.”

The locality for one specimen sent to the Carnegie Museum was given by Steinbach as Las Yuntas, 250 m., Dpto. Santa Cruz de la Sierra (fide C.J. McCoy, in litt.);9 this became the type of Aporophis melanocephalus, described by Griffin (1916), who failed to publish the department and who changed “Las Yuntas” to “Las Juntas.” The last is in fact the modern Bolivian name for at least one town previously with the former spelling, but there are at least two places with the name “Las Juntas” in the western part of the department of Santa Cruz,10 and we are uncertain which (if either) of them is the type locality of A. melanocephalus. The holotype of melanocephalus is a juvenile female only 185 mm in total length (tail  =  47 mm) rather than “291 mm” (tail 47 mm) as given by Griffin; the right maxilla has been subsequently removed and shows 25 + 2 teeth; there are about 137 ventrals and 69 subcaudals.

Viscera

In both species the tongue is long and extends back nearly (E. steinbachi) or quite (E. bassleri) to the heart. There is no left lung in the two specimens dissected and the trachea ends opposite the apex of the heart in both specimens.12 In each species the pulmonary (right) lung has the usual reticulation of raised alveolar rims on its lining (rather than the essentially smooth lining surface seen in a few colubrids such as Amastridium, Compsophis, and Psammodynastes); this alveolar reticulation is continued forward on the membranous dorsal wall of the trachea to fade into pitting and then into quite smooth membrane anteriorly, but in neither does this forward extension of alveolar reticulation bulge out beyond the dorsal tips of the cartilaginous tracheal semirings to form a conspicuous “tracheal lung.” In E. bassleri the forward extension of alveolar reticulation reaches within a head length of the head, but in E. steinbachi it reaches only about a heart length anterior to the heart. Thus, both species of Eutrachelophis seem to fall in the intermediate range, between Neotropical “xenodontines” that unequivocally have a tracheal lung (e.g., Coniophanes, Conophis, Darlingtonia, Hydrops, Rhadinaea decorata, R. laureata, Urotheca godmani) and those that unequivocally lack a tracheal lung (e.g., Atractus major, Ialtris, Taeniophallus brevirostris, T. occipitalis); probably the majority of “xenodontines” fall into the intermediate range with Eutrachelophis. The liver is separated from the heart by a moderate interval (eight ventrals in both specimens examined), as in most “xenodontines.” (But in a broad range of dipsadines, including Amastridium, Arrhyton vittatum, Darlingtonia, Rhadinaea calligaster, and Urotheca godmani, for example, the liver reaches forward nearly or quite to the level of the apex of the heart. At the opposite extreme, in Pseudoeryx plicatilis [AMNH R-52229] the liver is separated from the heart by 30 ventrals.)

As in other colubrids, with the exception of a few Old World genera (e.g., Boaedon, Bothrophthalmus, Liophidium, Pareas, Pseudoxyrhopus), the rectal caecum is absent in Eutrachelophis bassleri and E. steinbachi.

The general evidence from the viscera is consistent with a close affinity between Eutrachelophis bassleri and E. steinbachi, but cannot be considered convincing evidence for close relationship because the two species are not at all unusual in visceral features and the resemblances between them are shared with many other “xenodontines.” It is the head structure that shows a sufficient number of shared unusual characters to make a special common ancestry the most likely explanation for the resemblances between Eutrachelophis bassleri and E. steinbachi.

Head Glands

Both species show an unusually large temporal extension of the Harderian gland, exposed behind the orbit but posteriorly inserted deep to the muscle adductor mandibulae externus superficialis and superficial to the adductor externus profundus (sensu Zaher, 1994). An equally large (and similarly placed) temporal extension of the Harderian gland also occurs in Rhadinophanes monticola (UTA R-4176) and Contia tenuis (AMNH R-69062). In all these snakes, the deep insertion of the gland between adductores externus superficialis and profundus seems to form a functional complex, with the two muscles acting to compress and evacuate the gland; the profundus retains its usual function as a powerful adductor of the lower jaw, but the superficialis is reduced to a thin layer of fibers across the lateral surface of the rear portion of the gland and probably functions mostly—or entirely—as a compressor of the gland. Secretions from the Harderian gland discharge into the subbrillar space to lubricate the eye; the secretions also pass through the lachrymal duct into the vomeronasal (Jacobson's) organ, for a function still speculative (e.g., Bellairs and Boyd, 1947, 1950; Minucci et al., 1992; Rehorek et al., 2003: 358 [the last authors do not identify a method for lubricating the eyeball under the brille, and only observe that “there is no nictitating membrane in the orbit of either snake species”]).

In neither Eutrachelophis bassleri nor E. steinbachi could a rictal gland (sensu McDowell, 1986) be found, nor could any infolding of the oral mucosa just medial to quadrato-maxillary ligament that might represent a rictal gland; the quadrato-maxillary ligament ends anteriorly on the skin of the last supralabial, and so does not reach forward to the region where a rictal gland might be expected. McDowell (1986) argued that the “rictal gland” is the homolog of the “anterior temporal gland” and of at least the sheath of the venom gland of various other snakes, and that these are, in turn, homologous to the Mundplatte, or rictal fold, of lizards. Furthermore, he stated that since the lizard rictal fold is normally a long and deep invagination of oral mucosa deep to the quadrato-maxillary ligament, the absence of the rictal gland (and even of the portion of the ligament that should accompany it) in the two species of Eutrachelophis would be less lizardlike (and presumably, more specialized) than the presence of the gland; this absence would represent an unusual, but not unique, shared specialization. The gland is a minute vestige or absent in Taeniophallus brevirostris, and it could not be found in Conophis vittatus (AMNH R-65108), Contia tenuis (AMNH R-73392), Farancia abacura (AMNH R-110941), Helicops angulatus (AMNH R-52746), or Thamnodynastes pallidus (AMNH R-4446). In Urotheca multilineata (AMNH R-98288), at the opposite extreme, the rictal gland has become greatly expanded as a floccular, thin-walled glandular structure covering most of the temporal region just beneath the skin; although large, it is a solid mass of glandular tissue rather than a hollow pocket (thus, it is quite different in appearance from the lizard rictal fold) and may represent a secondary enlargement of the small to very small pocket seen in most “xenodontines.”

All three species of Eutrachelophis have a well-differentiated “supralabial gland,” the outline of which in some preserved specimens can easily be seen through the postorbital supralabial integument, as in figure 4 (unnamed species). The gland is similarly positioned in E. bassleri and E. steinbachi, in which it is adherent to the medial side of the supralabial integument. The serous (Duvernoy's) portion of this gland is not (to gross examination, at least) clearly differentiated from the mucous part of the gland in either species. Enlargement of the rear maxillary teeth is not accompanied by a correspondingly conspicuous differentiation of the gland, which tapers posteriorly rather than showing enlargement behind the level of the eye. Unfortunately, the serous and mucous contributions to this gland in Eutrachelophis and many other “xenodontines” are not distinguishable without histological preparation (Taub, 1966, 1967).

In both named species of Eutrachelophis the lateral nasal gland is well defined and lies in a clearly defined aditus conchae of the nasal capsule—that is, in an invagination of the lateral wall of the cartilaginous capsule that forms a vertically oriented protrusion (the concha) into the lateral wall of the nasal passage; the vertical edge of this protrusion defines a lateral diverticulum (the paracapsular recess, or sakter) of the nasal cavity, housed in the prefrontal bone, with the facial wing of the prefrontal forming a partial lateral wall for the paracapsular recess, the intraorbital wing of the prefrontal forming a posterior wall for the recess, and the roof of the lachrymal canal of the prefrontal forming the floor of the recess. Two bony processes define the rim of the aditus conchae: the conchal process of the septomaxilla rises along the anteromedial rim of the aditus; and the conchal process of the prefrontal, rising from dorsomedial rim of the anterior orifice of the lachrymal canal of the prefrontal, lies along the posterolateral rim of the aditus. The above conditions are as in the great majority of “xenodontines,” and Colubroidea in general. But some “xenodontines” have reduced the nasal conchae and its aditus, so that the bulk of the lateral nasal gland lies more superficially: in Tretanorhinus nigroluteus (AMNH R-70222) and in Coniophanes quinquevittatus (AMNH R-74493) the concha is present, but with a reduced aditus, so that much of the gland lies in a shallow depression on the side of the nasal capsule; in Apostolepis, Carphophis, Farancia, Hydrops, and Pseudoeryx, the concha (and its aditus) is only vaguely defined and the conchal process of the prefrontal is a blunt vestige or absent.

Head Muscles

Both Eutrachelophis bassleri and E. steinbachi have unusually weak jaw adductors, with the adductor externus medialis essentially confined to the lateral surface of the braincase, leaving a broad exposure of the parietal and supraoccipital between the left and right muscles.

The adductor externus superficialis (sensu McDowell, 1986  =  adductor externus profundus of most authors) and the adductor posterior are also weak muscles, so that the bony crests of the compound mandibular bone lateral (for the adductor externus superficialis) and medial (for the adductor posterior) to the mandibular adductor fossa are both low. These two mandibular adductors arise from the quadrate to insert on the mandible and therefore have the same mechanical force whether the quadrate is in the vertical (relaxed) position or is rotated outward (as when engulfing large prey); the weakness of these muscles, and of the mandibular crests for their insertion, indicates that neither species is well adapted to engulfing relatively large prey while exerting much force.

The conversion of the levator anguli oris into a compressor of the Harderian gland is mentioned above in discussion of the gland. Another unusual feature of this muscle shared by Eutrachelophis bassleri and E. steinbachi is that only the more posterior fibers of the usual colubrid levator anguli oris are retained; only those fibers that originate on the parietal are present and no fibers originate on the postorbital, nor do any fibers extend to the oral mucosa at the corner of the mouth or curve forward beneath the corner of the mouth. Diadophis punctatus (AMNH R-121701) and Liophis melanotus (AMNH R-98179) have a similar muscle, but usually the postorbital bears part of the origin of the levator anguli oris, with some of the anteriormost fibers bent forward around the corner of the mouth or inserted on the mucosa of the corner of the mouth. In some “xenodontines,” such as Atractus major (AMNH R-53782), Dipsas indica (AMNH R-53780), and Helicops angulatus (AMNH R-52746), these anterior recurrent fibers may be set off as a distinct muscle and loss of this distinct “1a” muscle would give the pattern seen in the two species of Eutrachelophis; however, there is no evidence to suggest that loss of the anterior fibers in bassleri and steinbachi was preceded by segregation of the fibers in a distinct muscle.

The pterygoideus in both species is of normal colubrid form, with both the pars major and pars minor joined anteriorly at their attachment to the anterolateral corner of the ectopterygoid (directly above the enlarged rear pair of maxillary teeth), but distinct at their posterior attachment to the mandible, the insertion of pterygoideus accessorius lying just behind the mandibular attachment of the pars minor.

In both Eutrachelophis bassleri and E. steinbachi, the retractor arcus palatini is unusually slender, its origin (from the sphenoid) being slightly narrower than the origin of the retractor vomeris (also on the sphenoid, immediately anteromedial to the origin of the retractor arcus palatini). In other colubrids examined, both New and Old World, the retractor arcus palatini is at least slightly larger (usually conspicuously larger) than the retractor vomeris. The insertion pattern of the retractor arcus palatini in both bassleri and steinbachi is the pattern that is most common among colubrids: fleshily, upon the posterior shaft of the palatine and, by tendon only, upon the choanal process of the palatine. The retractor vomeris and retractor arcus palatini have origins side by side on a nearly transverse (but arched forward anteriorly) muscular line on the sphenoid, and the protractor pterygoideus arises posterior to this muscular line; thus, the protractor pterygoideus lies behind the levator bulbi group (retractor vomeris and retractor arcus palatini), as in many “xenodontines” (e.g., Heterodon, Thamnodynastes), rather than extending forward on the sphenoid between the right and left levator bulbi groups (as in Atractus, Farancia, Helicops, Tretanorhinus, and many others).

Because the origins of the retractor vomeris and retractor arcus palatini muscles are arranged transversely across the sphenoid (rather than along the sphenoid-parietal contact, as they are in Diadophis, Farancia, Oxyrhopus, and many others), with the anterior orifice of the Vidian canal between the two muscles (as in Colubroidea generally), the anterior orifice of the Vidian canal is set well in from the border of the sphenoid (rather than near or on the sphenoid-parietal suture, as it is in Diadophis, Oxyrhopus, Tantalophis, and many others). The pattern of palatal muscles, both relative to one another and to the Vidian canal system of the sphenoid, is the same in both Eutrachelophis bassleri and E. steinbachi, but far from unique to them (for example, among genera related at the subfamilial level, the same pattern is seen in Thamnodynastes, but the pattern occurs also in the Madagascan Liopholidophis and Asiatic Pseudoxenodon and many other genera that probably are not closely related to the South American Eutrachelophis). This pattern seems to indicate only a moderate forward and backward movement of the palatine-pterygoid arch and very limited (if any) ability to rotate the palatine in the vertical plane. This is in keeping with the form of the palatine bone, with a long but slender choanal process closely applied to the vomer and a long shaft of the palatine extending posterior to the attachment of the palatine to the prefrontal. The opposite extreme would be seen in those genera with the origin of the protractor extending forward to the level of the eye and the origin of the retractor arcus palatini displaced far backward behind the level of the origin of the protractor; such a pattern allows considerably greater forward and backward movement of the palatine-pterygoid arch and is usually associated with loss of the choanal process of the palatine (as in Atractus) or with this choanal process becoming uncoupled from the vomer (as in Tretanorhinus, where the process is behind the choanae and attached by a long tether of connective tissue to the rear of the vomer).

In both Eutrachelophis bassleri and E. steinbachi, the protractor quadrati originates on the fascia covering the anterior end of the rectus capitis ventralis and there is no direct attachment to the basioccipital. In this, they resemble many other “xenodontines,” such as Conophis, Crisantophis, Hydrops, Liophis melanotus, Pseudoeryx, Rhadinaea decorata, and Thamnodynastes—but differ from some others, such as Adelphicos, Atractus, Dipsas indica, Heterodon, Rhadinaea flavilata, R. taeniata, Rhadinophanes, Taeniophallus brevirostris, and Tretanorhinus, where at least the most anterior fibers of the muscle are attached to the basioccipital. This seems to be a very unstable character of little taxonomic significance.

In both Eutrachelophis the cervicomandibularis has the usual colubrid insertion on the lower end of the quadrate (as a colubrid “retractor quadrati”); in both, the depressor mandibuli has a small but distinct occipital head, broadly separated from its fellow.

Except possibly for the narrowness of the retractor arcus palatini, none of the above features of the head musculature are unique to Eutrachelophis; overall, the head musculature of E. bassleri and E. steinbachi is much the same as in Liophis melanotus (AMNH R-98179) and many other “xenodontines.” What is most impressive, however, is the close agreement between bassleri and steinbachi in the muscular details that are known to vary even among other species that appear to be closely related. The unusual common features (15 scale rows and pale head and nuchal markings), which led to our comparing these two species in the first place, do not have any obvious functional correlation with details of the head musculature and it seems implausible that bassleri and steinbachi could bear such detailed correspondence in head musculature as a result of mere coincidence of random variations.

Skull and Dentition

The same argument just presented for head muscles, that concordance in characters known to be variable among “xenodontines” is too great to be accounted for by coincidence, has still greater force when considering the skulls of Eutrachelophis bassleri and E. steinbachi. To summarize in advance, the single skulls compared are so similar that coincidental resemblance must be rejected as an explanation.

What is most striking is that the skulls of bassleri and steinbachi combine an unusually short tabular, such as seen in some small-eyed burrowers (e.g., Atractus and Carphophis) with a construction of the orbit associated with large-eyed snakes (e.g., Dromicodryas, Psammodynastes, Thamnodynastes). When the details of foramina of the sphenoid and prootic are considered, there is a virtual identity between the skulls of Eutrachelophis bassleri and E. steinbachi, even though these foramina are so labile that closely related snakes may differ or, for that matter, even the left and right sides of the same skull may differ. Indeed, except for one character of the sphenoid in the orbital region (see pp. 32, 43), the single bassleri skull examined differs no more from the one steinbachi skull than might be expected for individual (including ontogenetic) variation within a single species.

This similarity is evident in the dentition, where the two skulls are nearly identical in form. In both species, the maxilla has a long series of small, evenly spaced teeth (x¯  =  25.6 in E. steinbachi, 27.4 in E. bassleri), followed by a diastema that is longer than the space occupied by a tooth socket, then two conspicuously enlarged, ungrooved teeth that are about twice as long as the prediastemal teeth; in both species, one of the enlarged pair of rear maxillary teeth is offset to the general line of the tooth row (fig. 12; see also fn. 1)—as is usually the case in “xenodontines,” but as noted in the discussion of head glands, there is no correspondingly conspicuous differentiation of Duvernoy's gland in either species to accompany the offset rear maxillary teeth. Eutrachelophis bassleri tends to have one or two more prediastemal maxillary teeth than the larger E. steinbachi. In both skulls, the palatine (with about 19–20 teeth) and pterygoid (about 30 teeth in bassleri, 34/36 in steinbachi) tooth rows form a continuous arcade of small teeth, longest on the anterior part of the palatine (where they are about equal to the adjacent maxillary teeth) and grading imperceptibly into much smaller teeth on the rear of the pterygoid. The last pterygoid tooth is approximately level with the occipital condyle in both skulls and the first palatine tooth lies just in advance of the orifice of the organ Jacobson and well behind the anterior end of the maxilla. The dentary teeth (34/32 in the bassleri skull, 35/33 in steinbachi) very gradually diminish posteriorly, but are small even at the anterior end of the bone (about equal to the prediastemal maxillary teeth). As usual in “xenodontines” (Apostolepis, Atractus, and Carphophis are among the exceptions), the last six (E. bassleri) or seven (E. steinbachi) dentary teeth are on a free posterior dentigerous limb of the dentary, behind the intramandibular hinge. None of the teeth is hinged at the base and all are of the usual pointed and recurved form.

The bones bearing the dentition are also nearly identical in E. bassleri and E. steinbachi. The maxilla has the form that is usual in colubrids, with a triangular anterior medial process that is applied to the ventral surface of the prefrontal and also is closely attached to the lateral edge of the lateral (maxillary) process of the palatine; this process of the palatine is also applied to the ventral surface of the prefrontal, just medial to the maxillo-prefrontal articulation. The maxilla has a well-developed posterior medial process with its apex, directed forwardly and medially, opposite the diastema anterior to the enlarged posterior maxillary teeth. The usual (for Colubroidea) ligament runs from the apex of this posterior medial process and the adjacent medial anterior process of the ectopterygoid to the apex of the anterior medial process of the maxilla and adjacent maxillary process of the palatine. As in most “xenodontines,” but unlike most proteroglyphs and a number of (mostly African) colubrids, such as Boaedon, Lycophidion, and Pseudaspis, this ligament is long and runs for most of the length of the orbit; this is not merely a consequence of a long maxilla, for although a short maxilla (e.g., Heterodon, Xenodon) will, of course, result in a short ligament, the converse is not true: Boaedon, for example, has a long maxilla but a short ligament because the posterior medial process of the maxilla (and adjacent medial anterior process of the ectopterygoid) lies well anterior to the rear of the maxilla, and the maxillary process of the palatine, which receives the anterior end of the ligament, lies well posterior to the prefrontal and to the tip of the anterior medial process of the maxilla (see appendix 2: fig. 39). In a number of snakes with a relatively short maxilla, such as Carphophis, the posterior medial process of the maxilla is extended backward, behind the tooth row, with its apex at the rear of the level of the orbit, so that the ligament (only feebly defined) is of moderate length.

The palatine is also very similar in Eutrachelophis bassleri and E. steinbachi. There is a long, narrow, transverse choanal process lying at nearly the exact middle of the palatine. This choanal process is positioned distinctly behind the level of the maxillary process and the ventral end of the prefrontal. The choanal process is entirely medial to the ventral end of the prefrontal, as usual in colubroids, in which no medial process of the prefrontal extends close to the frontal-septomaxillary joint dorsal to the palatine choanal process (unlike Dipsas indica [AMNH R-53780], Homalopsis, etc.) or functionally replaces the choanal process (as in Atractus).

The maxillary process has a similar triangular form in both Eutrachelophis bassleri and E. steinbachi, with a sphenopalatine canal for the nerve formed by fusion of the Vidian nerve with the infraorbital branch of the maxillary (V₂) nerve, as in many, probably most, “xenodontines.” The posterior shaft of the palatine (that is, the portion projecting behind the choanal process to meet the pterygoid) is unusually (but not uniquely) long and slender in both species and articulates with the pterygoid at the same transverse level as the ectopterygoid-maxillary articulation, rather than well anterior to the ectopterygoid-maxillary articulation (as in, for example: Amastridium veliferum, AMNH R-114309; Hydrops marti, AMNH R-52031; Manolepis putnami, AMNH R-58355; Oxyrhopus petola, AMNH R-52640; Rhadinaea decorata, AMNH R-107588; Taeniophallus brevirostris, AMNH R-15207) or well posterior to the ectopterygoid-maxillary articulation (as in, for example: Apostolepis flavotorquata, AMNH R-93559; Carphophis amoenus, AMNH R-75711). The articulation between the palatine and pterygoid involves a simple anterior end of the pterygoid (as usual in “xenodontines”) that is clasped by a short ventral lip of the palatine, bearing the last palatine tooth, and a longer dorsal finger of the palatine that extends back along the dorsal surface of the pterygoid; this is the most common palatine-pterygoid articulation among “xenodontines” (and colubrids in general), but other forms exist, such as simple end-to-end abutment (e.g., Heterodon, Apostolepis), or having the posterior finger of the palatine run along the medial, rather than dorsal, surface of the pterygoid (e.g., Farancia), or subequal posterior prongs of the palatine that fit, respectively, against the medial and the lateral surfaces of the pterygoid (e.g., Hydrops). The form of the palatine in both Eutrachelophis bassleri and E. steinbachi is very similar to that of Rhadinaea decorata (AMNH R-107588), R. fulvivittis (AMNH R-100890), and R. taeniata (AMNH R-106933), but quite different from that of Taeniophallus brevirostris (AMNH R-15207), where the choanal process has a broad base extending back almost to the level of the prongs for the pterygoid, or Urotheca multilineata (AMNH R-98288), where the slender choanal process has a distinct forward hooking of its apex and the shaft behind it is short.

Eutrachelophis bassleri and E. steinbachi have very similar pterygoid bones, as shown in the figures, but this is of less weight than the similarities in the palatine bone because differences in pterygoid bone shape are minor and subtle in “xenodontines” generally. The similarity in shape of the ectopteryoid between E. bassleri and E. steinbachi is more impressive, because this bone shows a wide range of form within the Xenodontinae. In both, the bone is moderately long (but conspicuously shorter than the maxilla), with a curved and cylindroid shaft that is free of the pterygoid for more than half its length and is not expanded in its posterior portion that is applied against the dorsal surface of the pterygoid. In both, the anterior end is abruptly expanded and flattened and is very asymmetrically divided, by a broadly rounded anterior emargination, into an acutely triangular medial anterior process and a nearly rectangular lateral anterior process; the medial anterior process extends conspicuously anterior to the lateral anterior process. Thus, the two species differ from, for example, Carphophis, Contia, and Farancia, in which the medial anterior process is greatly reduced; they also differ from such “xenodontines” as Conophis vittatus (AMNH R-65108) and Manolepis putnami (AMNH R-58355), in which the medial anterior process is broader—and longer—than the lateral anterior process. Downs (1967) has documented the considerable range in ectopterygoid form within the genus Geophis, where the form of the bone is distinctive of species groups within the genus. This seems to be true also of Atractus: some species (e.g., A. elaps, AMNH R-28843) have lost the lateral anterior process of the ectopterygoid but retain a long free shaft of the bone; others (e.g., A. major, AMNH R-53782) retain anterior furcation of the bone but have so shortened the free shaft that the maxilla is probably immovable relative to the pterygoid; and still others (e.g., A. trilineatus, AMNH R-101336) retain a long free shaft and also anterior furcation of the ectopterygoid. This variation within a single genus suggests that the form of the ectopterygoid is easily modified and that differences between two species might be expected even if the species are closely related; however, in the case of E. bassleri and E. steinbachi it is the great similarity between the two species that must be accounted for, and close phyletic relationship seems the simplest explanation.

The great similarity in the compound bone of the mandible has been noted in the discussion of the head muscles. It may be added that both species agree in having a long (for Colubroidea) retroarticular process and in the details of the splenial-angular-dentary complex. In both, the Meckelian canal is open anterior to the splenial nearly to the anterior end of the dentary, as in, for example, Amastridium, Rhadinophanes, and Coniophanes fissidens (AMNH R-69977). However, in such “xenodontines” as Coniophanes quinquevittatus (AMNH R-74493), Rhadinaea laureata (AMNH R- 68380), and Tantalophis, the lips of the dentary forming the dorsal and ventral edges of the Meckelian exposure meet, but leave a suture, and in Liophis melanotus (AMNH R-98179), Farancia abacura (AMNH R-110941), and Apostolepis flavotorquata (AMNH R-93559) this suture fuses, at least anteriorly.

The splenial and angular are equal and moderately long; each of the bones contains a mylohyoid foramen; the splenial extends about halfway forward from the splenial-angular hinge articulation to the anterior end of the dentary and the dorsal edge of the splenial is separated by a fissure from the dentary (and so, the Meckelian canal is narrowly open, even opposite the splenial). This is as in many other “xenodontines,” but in some others (e.g., Hydrops marti, AMNH R-52031; Oxyrhopus petola, AMNH R-52640) the splenial is smaller, relative to the dentary, and falls well short of the halfway point between the splenial-angular hinge and the tip of the dentary; in Apostolepis, Hydrops, and Thamnodynastes, the splenial is distinctly shorter than the angular; variations in the opposite direction are seen in Coniophanes imperialis (AMNH R-77064), whose splenial is conspicuously longer than the angular, and in Farancia the splenial is unusually large relative to the dentary (but subequal in length to the angular) and extends well anterior to the midpoint between the splenial-angular hinge and the tip of the dentary. In both Eutrachelophis bassleri and E. steinbachi, the splenial has a narrow, fingerlike process that extends upward along the anterior edge of the angular; excluding that bone from the rim of the Meckelian exposure; this posterior dorsal process of the splenial is present and sharply defined in many other “xenodontines” (e.g., Carphophis amoenus, AMNH R-121650; Farancia abacura, AMNH R-110941; Oxyrhopus petola, AMNH R-52640; Rhadinaea decorata, AMNH R-107588; Urotheca multilineata, AMNH R-98288), but the process is absent in many others (e.g., Amastridium veliferum, AMNH R-114309; Coniophanes imperialis, AMNH R-77064; Conophis vittatus, AMNH R-65108; Hydrops marti, AMNH R-52031; Pseudoeryx plicatilis, AMNH R-52229; Thamnodynastes pallidus, AMNH R-4446). It should be noted that this tiny sliver of the splenial bone is probably of more functional importance to the feeding mechanism than its size would suggest; the posterior dorsal process of the splenial lies immediately dorsal to the condyle-cotyle articulation between the splenial and angular, a circular universal joint, the edge-to-edge contact of the posterior dorsal process of the splenial with the anterior border of the angular limits the plane of rotation of the hinge. In most “xenodontines” that have this process of the splenial, including Eutrachelophis bassleri and E. steinbachi, this edge-to-edge contact is diagonal to the long axis of the jaw, and so imparts a rotation of the dentary around the long axis when the intramandibular hinge is flexed (in Farancia, where the edge-to-edge contact is vertical, no such rotation around the long axis is permitted).

The construction of the orbit in both Eutrachelophis bassleri and E. steinbachi shows a pattern that has developed independently among natricines (e.g., Psammodynastes, Rhabdophis miniatus), pseudoxyrhophiines (e.g., Dromicodryas, Thamnosophis lateralis but not Liopholidophis sexlineatus), and other colubrids in association with a large eye and small olfactory forebrain. It may be considered a consequence of secondary enlargement of the eye, but it is not a necessary or automatic consequence, since such notably large-eyed snakes as Boiga do not show this pattern; that Leioheterodon, a relatively small-eyed Madagascar snake, shows many features of the pattern can perhaps be explained as the result of derivation from a large-eyed ancestor similar to Dromicodryas and Thamnosophis lateralis. The conspicuous features of the pattern are:13

Of the features of the pattern just listed, Eutrachelophis bassleri does not show number 5 above; the [para]sphenoid rostrum is flat dorsally. In E. steinbachi there is a distinct crest on the dorsal side of the [para]sphenoid rostrum, but the crest is less developed than in Conophis, Crisantophis, Manolepis, and Thamnodynastes, and meets the frontals only anteriorly, leaving a triangular open chink between the [para]sphenoid crest and the frontals more posteriorly. Except for the better development of the keel on the sphenoid and the absence of a fenestra in the trabecular groove of the sphenoid, Thamnodynastes pallidus (AMNH R-4446) closely resembles E. bassleri and E. steinbachi in the construction of the orbit, but Conophis and Manolepis (these points are uncertain for Crisantophis, observed by dissection only, on AMNH R-112402) lack fenestration between the frontal and parietal, have only feeble suborbital laminae of the [para]sphenoid, and have a narrower entry of the parietal into the orbital rim, but they do have a perforation in the trabecular groove. Liophis melanotus (AMNH R-98179) approaches the pattern of E. bassleri and E. steinbachi in a different way: there is a similar fenestration between frontal and parietal, similar development of the [para]sphenoid suborbital laminae, and a similarly broad entry of the parietal into the orbital rim, but the forebrain cavity of the frontals is only slightly raised above the trabeculae posteriorly (nevertheless, the sphenoid has a low but quite distinct dorsal crest to meet the frontals here, and there is no open chink between frontals and sphenoid such as is seen in E. steinbachi); anteriorly, the frontals rest upon the trabeculae and form weak supratrabecular ridges. A third, slighter approach to the orbital pattern of E. bassleri and E. steinbachi is made by Taeniophallus brevirostris (AMNH R-15207): the forebrain chamber is only slightly elevated above the trabeculae, but, nevertheless, the frontals are narrowly—but completely—excluded from the trabecular grooves by a narrow dorsal lip of the trabecular groove of [para]sphenoid; the parietal rather broadly excludes the frontal from the postorbital, but otherwise the orbit is quite different from that of E. bassleri and E. steinbachi, without frontal-parietal fenestration and with only feeble suborbital flanges of the [para]sphenoid.

Most Rhadinaea, sensu lato, such as R. decorata (AMNH R-107588), R. flavilata (AMNH R-50491), R. laureata (AMNH R-68380), R. taeniata (AMNH R-106933), and Urotheca multilineata (AMNH R-98288), have frontals that rest upon the trabecular cartilages and form strong supratrabecular crests that even descend, lateral to the trabecula, to meet the ventral lip of the [para]sphenoid groove for trabecula at the front of the orbit. Many other “xenodontines” (and other colubroids) show this pattern, such as Amastridium, Coniophanes fissidens (AMNH R-69977), C. imperialis (AMNH R-77064), Rhadinophanes, and Tantalophis; or the contact of the frontal with the [para]sphenoid lateral to the trabecular cartilage may be much longer and extend for most or all of the length of the frontal supratrabecular crest as in Coniophanes quinquevittatus (AMNH R-74493, by dissection), Adelphicos, Atractus, and Tretanorhinus. Such long contact of the frontal with the trabecula can lead to contact of the supratrabecular crest of the frontal with ventral end of the parietal beneath the orbital foramen, excluding the sphenoid from that foramen, as in Nothopsis and Hydrops. This is the usual condition in noncolubroid snakes and in many proteroglyphs and seems to be associated with a relatively small eye, at least relative to the size of the forebrain and snout. In all forms showing this “small-eyed pattern” the suborbital lamina of the [para]sphenoid is absent and there is no fenestration between the frontal and parietal in the medial wall of the orbit. The same may be said for those “xenodontines” with the “moderate-eyed pattern.” such as Coniophanes and Rhadinaea, except that a slight suborbital flange of the [para]sphenoid may be present.

There seems, then, to be a series from the “small-eyed pattern.” as seen in such “xenodontines” as Hydrops, to the “large-eyed pattern” as seen in such “xenodontines” as Thamnodynastes. The series will not describe all the intermediates precisely, since there can be extensive fenestration of the wall of the orbit combined with the forebrain cavity resting on the trabeculae (as in Liophis melanotus) or lack of this fenestration in combination with high elevation of the forebrain cavity above the trabeculae (as in Manolepis putnami). Heterodon is hard to place, since it has enormous suborbital flanges of the [para]sphenoid and there is fenestration of the cranial-orbital wall (but mainly by retraction of the anterior edge of the parietal); however, the forebrain cavity seems to have been secondarily depressed toward the trabeculae (but does not touch them), so that the crest on the [para]sphenoid is concealed from external view by the frontals. Because of its long snout, Heterodon has only a “moderately” large eye relative to total head length, but the orbit is large relative to the unusually short braincase. In spite of these complications in surveying all “xenodontines,” the central point of these comparisons remains: Eutrachelophis bassleri and E. steinbachi are extremely similar to each other in orbit construction, and however the series from “small-eyed pattern” to “large-eyed pattern” is arranged, the two species under special consideration are at nearly the same step in the series, near the “large-eyed pattern” extreme; this in spite of the fact that neither species has exceptionally large eyes. However, the only significant skull difference between the two species occurs in the orbital region—Eutrachelophis steinbachi has a dorsal crest of the parasphenoid rostrum that is exposed below the frontals and E. bassleri does not.

Entry of the parietal into the orbital border occurs in Amastridium, Taeniophallus brevirostris (AMNH R-15207), and some Rhadinaea (R. decorata, AMNH R-107588; R. taeniata, AMNH R-106933), but most Rhadinaea, like Adelphicos, Atractus, Coniophanes, Tantalophis, Rhadinophanes, and Tretanorhinus, have a postorbital-frontal contact excluding the parietal from the rim of the orbit. In Apostolepis (notably small-eyed snakes), the parietal enters the orbit extensively, nearly (A. flavotorquata, AMNH R-93559) or very much (A. erythronotus, AMNH R-62192) excluding the frontal. Parietal entry into the orbital rim is thus not a simple function of orbit size. This increases the significance of parietal entry into the orbital rim in both Eutrachelophis bassleri and E. steinbachi as a special resemblance between the two species and not a mere duplication of another character.

Eutrachelophis bassleri and E. steinbachi have very similar prefrontal bones that are unusually short anteroposteriorly, so that the vertical (frontal-to-maxillary) height is more than twice the anteroposterior dimension; this anteroposterior shortness largely reflects the weak development of the facial wing of the bone that extends forward over the lateral surface of the paracapsular (sakter) region of the cartilaginous nasal capsule. The facial wing of both E. bassleri and E. steinbachi lacks the forward extension and dorsad hooking (apomorphies of Pseudoboini) seen in Drepanoides anomalus, AMNH R-53419, and Oxyrhopus petola, AMNH R-52640; moreover, it does not cover the preorbital area as in Hypsirhynchus ferox, AMNH R-40124, and Manolepis putnami, AMNH R-58355. However, the facial wing is small and obtusely pointed, with the result that the prefrontal does not have the simple and broadly convex anterior border seen in Helicops angulatus, AMNH R-56031, or the straight and vertical anterior border seen in Farancia abacura, AMNH R-110941. The attachment of the prefrontal to the frontal is as in most other “xenodontines” (and the great majority of colubrids): dorsally and superficially, there is a tongue-in-groove articulation, with the edge of the prefrontal fitting into a groove on the frontal; deep and ventral to this, there is a squamous overlap, with the intraorbital wing of the prefrontal fitting just behind a transversely oriented lateral eversion of the part of frontal that forms the lateral rim of the olfactory foramen (this lateral eversion of the frontal lies against the rear wall of the olfactory capsule, thus partially separating the intraorbital wing of the prefrontal from the capsule, and prevents forward rotation of the prefrontal). In both Eutrachelophis bassleri and E. steinbachi, the superficial tongue-in-groove articulation is about 30° from transverse in its orientation relative to the axis of the skull, with the prefrontal extended along the anterior border of the frontal (but falling far short of the midline or of the nasal and permitting broad entry of the frontal into the rim of the dorsal exposure of the nasal capsule); there is no suggestion of posterior extension of the prefrontal along the lateral (orbital) border of the frontal. This is as in some other “xenodontines” such as Liophis melanotus (AMNH R-98179), Coniophanes fissidens (AMNH R-69977), and C. imperialis (AMNH R-77064), but the majority have the tongue-in-groove articulation oriented at 45° and in several (e.g., Carphophis amoenus, AMNH R-121650; Conophis vittatus, AMNH R-65108; Manolepis putnami, AMNH R-58355; Pseudoeryx plicatilis, AMNH R-52229) it is nearly longitudinal (but without backward extension of the prefrontal over the eye). In Hydrops marti (AMNH R-52031), the tongue-in-groove articulation is perfectly longitudinal and the deeper squamous articulation is absent (the lateral eversion of the frontal is missing), so that some transverse rotation of the prefrontal probably is permitted.

In both Eutrachelophis bassleri and E. steinbachi, the facet of the parietal bone bearing the postorbital bone is moderately long and oriented diagonally, approaching a vertical orientation. This is as in most “xenodontines” (e.g., Alsophis, Liophis, Manolepis, Rhadinaea decorata, Taeniophallus brevirostris, Urotheca multilineata, and Thamnodynastes) and imparts an anteroposterior motion to the tip of the postorbital when that bone rotates; since the tip of the postorbital is bound by ligament to a low but distinct elevation on the dorsal edge of the maxilla (just in front of the maxillary diastema in E. bassleri and E. steinbachi and most of the others), the angle of attachment of the postorbital to the parietal probably has an indirect control over the movements of the rear of the maxilla. In some “xenodontines” (e.g., Atractus) the attachment of the postorbital to the parietal is so short that it is a virtual pivot joint, and in others (e.g., Apostolepis and, according to Downs, 1967, some Geophis) the postorbital is absent, so that the maxillary is tethered to the braincase only by a long and quite flexible ligament; this seems to be associated with a short maxilla, but is not an “automatic” consequence of a short maxilla, since Heterodon and Xenodon (see Kardong, 1979), both with an unusually short maxilla (but a maxilla that rotates strongly in a vertical plane), have a more or less vertical hinge attachment of the postorbital to the parietal. In some “xenodontines” (e.g., Adelphicos, Carphophis, Contia, Diadophis, Farancia, Hydrops, Pseudoeryx) there is a long postorbital-parietal hinge, but oriented nearly or quite horizontally, so that rotation of the postorbital is transverse; probably this acts, through the maxillary-postorbital ligamentous connection, primarily as a check preventing excess lateral displacement of the rear of the maxilla. Still others (e.g., Amastridium, Coniophanes, Oxyrhopus, Rhadinaea flavilata) are intermediate, with the attachment of the postorbital closer to the horizontal than in Eutrachelophis bassleri and E. steinbachi, but nonetheless distinctly inclined.

In both Eutrachelophis bassleri and E. steinbachi, the rostral complex of bones (premaxilla, nasals, septomaxillae, and vomers) is very similar and the rostral complex in both is small relative to the skull. The premaxilla has quite distinct lateral processes, but the bone is rather small (about as in Liophis melanotus, AMNH R-98179). As in Amastridium, Farancia, Hydrops, Liophis, Pseudoeryx, Rhadinaea decorata, Urotheca multilineata, and others, the premaxilla and vomer make an overlapping contact, so that the vomer participates in the support of the premaxilla (but, as is true of Colubroidea in general, is less important than the septomaxilla in this support); in many other “xenodontines” (e.g., Adelphicos, Apostolepis, Atractus, Carphophis, Conophis, Heterodon, Manolepis, Oxyrhopus, Rhadinaea flavilata, R. laureata, R. taeniata), the vomer makes only a point-to-point contact with the premaxilla or is separated from that bone. In both Eutrachelophis bassleri and E. steinbachi, there is a large intervestibular fenestra, between the narial vestibules of the opposite sides and bounded anteriorly by the ascending process of the premaxilla, ventrally by the septomaxillae, and posterodorsally by the nasals; a similar intervestibular fenestra is seen in many other “xenodontines” (e.g., Adelphicos veraepacis, Liophis melanotus, Rhadinaea taeniata, Thamnodynastes pallidus), but in many others (e.g., Amastridium, Apostolepis, Carphophis, Farancia, Hydrops, Rhadinaea decorata, R. laureata, Urotheca multilineata) the intervestibular fenestra is small or absent.

Both Eutrachelophis bassleri and E. steinbachi have the usual form of hinge between the rostral complex and the frontals for “xenodontines”: Along the cartilaginous nasal septum the septomaxilla sends back a posterior process, whose posterior tip bends strongly outward behind the passage for the vomeronasal portion of the olfactory nerve. The lateral flexure of the posterior process of the septomaxilla forms a posteriorly convex facet for the corresponding septomaxillary facet of the frontal that, in turn, is well defined by a distinct peduncle facing forward and slightly but distinctly mediad; the nasals touch the interolfactory pillar of the frontals, but without forming any articular surface, between the frontal-septomaxillary articulations and do not make any contact with the septomaxillary facets. Although most “xenodontines” (and other colubrids) have a similar hinge between the rostral complex and the frontals, there are some exceptions. Thus, in Apostolepis, the posterior process of the septomaxilla is stout and abuts end to end with the frontal facet; in Carphophis, the nasals are expanded and fused at their contact with the interolfactory pillar of the frontals, forming an articulation addition to the septomaxillary-frontal articulation; in Pseudoeryx and Hydrops, the nasals do not reach the frontals, nor is there any nasal-frontal contact in Atractus crassicaudatus (AMNH R-24235) or Adelphicos veraepacis (AMNH R-66961).

In both Eutrachelophis bassleri and E. steinbachi, the tip of the [para]sphenoid, just ventral to the fusion of the paired trabeculae that forms a trabecula communis, is distinctly expanded laterally and shallowly trilobated by a pair of obtuse notches. This is true also of Coniophanes, Farancia, Rhadinaea flavilata, R. laureata, R. taeniata, and many other “xenodontines.” It is seen also, so far as the tip of the parasphenoid is concerned, in Atractus, but in that genus the entire anterior end of the parasphenoid is broad, as in Adelphicos, Ninia, Tretanorhinus, and others, and so the tip of the parasphenoid does not appear broad relative to the interorbital portion. In others, such as Rhadinophanes and Tantalophis, as well as Adelphicos, Ninia, etc., the apex of the parasphenoid is convexly rounded anteriorly; in Amastridium and Thamnodynastes pallidus, and others, the tip of the parasphenoid is simply forked by a shallow median emargination; in Apostolepis, Carphophis, Conophis vittatus, Contia, Diadophis, Heterodon, Manolepis, and Nothopsis, the parasphenoid is pointed anteriorly.

The suborbital flanges of the parasphenoid have been noted in the preceding comparisons of the orbit; as is probably true of all snakes with a parasphenoid that is broadened to form suborbital flanges, the parasphenoid of both Eutrachelophis bassleri and E. steinbachi rises to meet the parietal lateral to the trabecular cartilages, so that the ossified bases of the trabeculae (radices trabeculae) are concealed intracranially; this is the usual condition in “xenodontines” (even in genera such as Atractus, Hydrops, and Pseudoeryx, with little or no development of suborbital flanges), but in some (e.g., Apostolepis, Carphophis, Farancia), the parasphenoid appears to be narrower in this region, failing to surround the radices trabeculae laterally, and so the radices trabecular are exposed externally, lateral to the contact of the sphenoid with the more anterior part of the parietal.

The parasphenoid of both Eutrachelophis bassleri and E. steinbachi seems somewhat shorter posteriorly than in the majority of “xenodontines,” since it leaves the foramen for entry of the carotid artery into the pituitary fossa separate from the more lateral foramen for entry of the palatine ramus of the facial nerve into the rear of the Vidian canal. This is seen also in Liophis melanotus (AMNH R-98179) and Urotheca multilineata (AMNH R-98288), as well as in many non-“xenodontines,” but usually the parasphenoid of “xenodontines” extends back beneath these two foramina to define a common palatine nerve–carotid artery canal that forks within the sphenoid into a carotid and a Vidian canal. (In Heterodon, there is an anomalous condition, in which the posterior end of the Vidian canal is in the prootic and the single foramen in the rear of sphenoid is for the carotid artery alone. Atractus and Apostolepis also appear to have the posterior orifice of the Vidian canal within the chamber of the prootic for the trigeminal ganglion, so that only the carotid enters the large foramen in the posterior part of the sphenoid; in these forms it is probably impossible to guess the precise limits of the parasphenoid relative to the perichondral sphenoid posteriorly.) The relationships of the Vidian canal system of the sphenoid to the protractor pterygoideus and levator bulbi group of muscles have been discussed in connection with those muscles, but it may be added here that the bony canal for the palatine nerve is unusually short, as compared with other “xenodontines,” in both Eutrachelophis bassleri and E. steinbachi.

In both species, there is a rather large foramen (pv) in the lateral edge of the sphenoid, immediately adjacent to the sphenoid-parietal suture and well anterior to the sphenoid-prootic contact (boldface abbreviations indicate foramina shown in figs.10–11). This foramen, presumably for the pituitary vein and for the entry of the retractor pterygoideus ramus of the trigeminal nerve (V₄ levator bulbi) into the cranium is larger than that of any Rhadinaea examined, or that of Amastridium, Coniophanes, Liophis melanotus, Rhadinophanes, and Tantalophis, but similarly placed; Farancia and Thamnodynastes show close agreement with Eutrachelophis bassleri and E. steinbachi both in size and position of the foramen; in Pseudoeryx the foramen is large but more posteriorly placed, just anterior to the sphenoid-prootic contact, and in Oxyrhopus petola (AMNH R-52640) and Hydrops marti (AMNH R-52031) the foramen lies in the parietal-prootic suture. In both Eutrachelophis bassleri and E. steinbachi, the foramen on the left side appears to be as large as that on the right side, probably indicating that the drainage from the pituitary and middle cerebral veins through this foramen is approximately equal on the left and right sides, but in Carphophis amoenus (AMNH R-75711), Contia tenuis (AMNH R-69062), and Rhadinaea flavilata (AMNH R-50491) the right pituitary vein foramen is conspicuously larger than the left, a frequent asymmetry in colubroids related to a tendency to have blood enter the pituitary on the left (from the functional carotid) and leave on the right, through a large pituitary vein.

Eutrachelophis bassleri and E. steinbachi have very similar prootic bones, including the pattern of foramina in the region of the trigeminal and facial foramina. This pattern is imposed by the fusion of the alethinophidian bridge, or “Gaupp's bone,” to the outer side of the prootic—external to branching of the trigeminal and to the petrosal sinus or network of veins around the trigeminal ganglion. The precise pattern of foramina depends on just where the edges of the alethinophidian bridge lie relative to a complex pattern of veins and nerves that cross one another. The pattern appears to be the same in the single skull each of Eutrachelophis bassleri and E. steinbachi, a remarkable coincidence if they are not closely related, particularly given that it was impossible to match this pattern with any of the other “xenodontine” (or other colubrid) skulls compared. In both Eutrachelophis bassleri and E. steinbachi, foramen V₂ for the maxillary nerve is broadly separated from foramen V₃VII for the mandibular and facial nerves by the more dorsal portion of the alethinophidian bridge, as in Apostolepis, Oxyrhopus, Rhadinaea, and others; in Hydrops, Manolepis, Tantalophis, and others, the two foramina are closer together, and in some Heterodon (e.g., H. nasicus, AMNH R-109431) the two foramina are confluent because of failure of the dorsal part of the alethinophidian bridge to extend between V₂ and V₃. Dorsal to the aperture for V₃, both have a small foramen opening posterodorsally, probably for the vein connecting the petrosal sinus with the main stem of the vena capitis lateralis. Urotheca multilineata (AMNH R-98288) has a similar foramen, but also another, probably venous, foramen just posterodorsal to the aperture for V₂. A venous (?) foramen posterodorsal to the V₂ aperture, but no venous foramen near the V₃ aperture, was found in Rhadinaea laureata (AMNH R-68380) and Thamnodynastes pallidus (AMNH R-4446); most of the “xenodontines” compared had no venous foramina dorsal to the apertures for V₂ and V₃. Ventral to the aperture for V₃, both Eutrachelophis bassleri and E. steinbachi have a small foramen facing downward and forward, probably for the palatine ramus of the facial nerve (VII pal.) just anterior to this foramen (and sharing a common longitudinal groove with it) is a small foramen facing downward and backward. To judge from dissection of Coniophanes quinquevittatus (AMNH R-76693) this foramen is for the small trigeminal ramus to the protractor quadrati muscle (V₄ pro. quad.); dissection of Tantalophis discolor (AMNH R-103130) revealed two similarly placed foramina, but with the VII pal. foramen anterior to the V₄ pro. quad. foramen; since the two nerves cross in this region, just which foramen is the more anterior depends on precisely how far ventrally (i.e., to just above or to just below the crossing) the alethinophidian bridge extends. The direction the foramina face indicates that Eutrachelophis bassleri and E. steinbachi are more like Coniophanes than like Tantalophis in this aspect of the pattern.

Ventral to foramen V₂, Eutrachelophis steinbachi has a rather large foramen, but E. bassleri has a slightly larger foramen with a conspicuously smaller foramen immediately behind it; probably the larger foramen of E. bassleri is for the venous connection of the pituitary–middle cerebral vein to the petrosal sinus and the smaller foramen immediately behind it is for the retractor pterygoideus nerve (V₄ levator bulbi). The confluence of the two foramina in Eutrachelophis steinbachi is by far the more usual condition. Most “xenodontines” (apart from the anomalous Heterodon, whose foramina associated with the alethinophidian bridge are difficult to homologize with other snakes) have a V₄ levator bulbi foramen similar to that of Eutrachelophis steinbachi, but in Urotheca multilineata (AMNH R-98288) and Rhadinophanes the foramen is very small, suggesting it is for the nerve alone and the vein is absent, and, to judge from the size of the foramen, the vein must have been very small in Rhadinaea flavilata (AMNH R-50491) and Coniophanes fissidens (AMNH R-69977). In Pseudoeryx plicatilis (AMNH R-52229) the V₄ levator bulbi foramen is confluent with the aperture for V₂ and in Hydrops marti (AMNH R-52031) the V₄ levator bulbi foramen is anterior to, rather than below, the V₂ aperture.

In summary, the presence of venous (?) foramina dorsal to the trigeminal ganglion, together with the details of the pituitary vein foramen, contribute to the unusual pattern of the two Eutrachelophis skulls examined.

As in most “xenodontines,” both Eutrachelophis bassleri and E. steinbachi have a large and longitudinally oval footplate of the stapes that is overhung by a dorsal crista circumfenestralis, formed about equally by the prootic and exoccipital, and partially covered from below by the longissimus crest of the exoccipital, which conceals the recessus scalae tympani from lateral view. Although the skull of each was prepared from a male with well-mineralized hemipenial spines and with a strongly convoluted vas deferens (thus, presumably mature), in neither species does the dorsal crista circumfenestralis touch the longissimus crest to enclose the stapedial footplate entirely (such complete enclosure is uncommon in “xenodontines,” but occurs in Apostolepis flavotorquata, AMNH R-93559, Arrhyton taeniatum [see Maglio, 1970], Taeniophallus brevirostris, AMNH R-15207, and in Urotheca multilineata, AMNH R-98288). At the opposite extreme is Heterodon, where the stapes is almost entirely exposed to direct lateral view.

In both Eutrachelophis bassleri and E. steinbachi, the tabular (supratemporal or squamosal of some authors) is small and barely projects posteriorly beyond the sheath (in the exoccipital bone) of the posterior semicircular canal; anteriorly, it falls well short of the parietal bone, but, nevertheless, it is exposed anterior to its contact with the quadrate and is about equal in length to the quadrate. Thus, the reduction of the tabular, which results in the suspension of the quadrate lying opposite, rather than behind, the rear of the otic capsule, is more as in Apostolepis, Atractus, and Carphophis, than as in Dipsas and Sibon. As in other “xenodontines” (but unlike genera such as Boaedon, Elapoidis, Lycophidion, and Psammodynastes), the lateral margin of the tabular is gently sinuate, without a distinct and angular lateral lobe extending onto the dorsal crista circumfenestralis. Further, as in most “xenodontines” (even those, such as Oxyrhopus and Thamnodynastes, with a strong backward extension of the tabular), there is no tubercle or ridge or countersinking of the cranial surface for the tabular and that bone would seem to have a slight mobility in its cranial attachment. Hence, since the quadrate makes no direct contact with the cranium, some slight adjustment in the angle of rotation of the quadrate (relative to the cranium) seems possible, although the lack of backward projection of the tabular would make the leverage for this adjustment very small in the case of Eutrachelophis bassleri and E. steinbachi. In genera with an angular notch in the lateral border of the tabular (such as Psammodynastes and Lycophidion), a tubercle of the prootic fits into this notch and locks the tabular into position. In contrast, the xenodontine Oxyrhopus (as noted during preparation of the skull of O. petola, AMNH R-52640), although having a long tabular projecting well behind its contact with the cranium, has the tabular held in position by “guy wires;” that is, strong ligaments, one from the nuchal crest of the supraoccipital to the medial edge of the tabular and another from a longitudinal ridge on the prootic to the lateral edge of the tabular, hold the tabular in place. A few “xenodontines” seem to have a rigidly locked tabular: in Heterodon platyrhinos (AMNH R-63590), the surface of the prootic for the tabular is corrugated, and in Apostolepis flavotorquata (AMNH R-93559), the tabular is strongly S-shaped, with an irregular outline that fits a countersunk depression on the exoccipital behind the crista circumfenestralis.

The quadrate is of similar triangular form in Eutrachelophis bassleri and E. steinbachi, but less backswept in the latter; in both, the stapedial facet (processus internus module) is at almost the exact center of the posterior profile of the bone, rather than conspicuously ventral to the center (as in Farancia) or distinctly above the center (as in Rhadinaea decorata, AMNH R-107588).

Both Eutrachelophis bassleri and E. steinbachi have the most common form of occipital condyle among “xenodontines” and other colubrids: conspicuously narrower than the foramen magnum, but transversely oval in form, with the exoccipitals well separated by the basioccipital, and without distinct peduncle or neck; in some “xenodontines,” such as Farancia abacura (AMNH R-110941) and Carphophis amoenus (AMNH R-121650) the condyle may be much broader, as broad as the foramen magnum; in others, such as Apostolepis flavotorquata (AMNH R-93559) and Pseudoeryx plicatilis (AMNH R-52229), the condyle is distinctly pedunculate and the exoccipitals meet above the basioccipital.

Color Pattern

The color pattern of snakes can be as useful in phylogenetic studies as it is in identification, but authors tend to yield precedence when conflicting hemipenial data are involved. For example, the sister genera Pliocercus and Urotheca (the latter equivalent to the former lateristriga group of Rhadinaea) share the characters of a long, disproportionately thick tail and a hemipenis having a small, naked pocket in the asulcate edge of the capitulum (Myers, 1974: 273, fig. 4). On this basis, Savage and Crother (1989) united the two genera. Myers and Cadle (1994: 3) disagreed with that action, saying that:

One wonders whether “mimicry” of a sort might also have been involved in the evolutionary development of the distinctive color pattern of Eutrachelophis. This pattern includes striking dark-rimmed whitish ocelli or elongated spots on the head, or head and neck, with a weak semblance of dorsal dark spots/stripes anteriorly, becoming nearly uniform posteriorly. See figures 1–2 and 4–Fig. 5.Fig. 6.Fig. 7.8 in the species accounts. There is both inter- and intraspecific variation in the alignment of the pale ocellar markings, but the variation seems relatively minor and the combined patterns seem to us diagnostic at the generic level. Eutrachelophis bassleri and E. steinbachi differ from one another in the alignment of their postocular and nuchal markings (compare figs. 2 and 8)—but this minor variation is in total greatly exceeded in the similar markings of Rhadinaea decorata, as portrayed in Myers (1974: fig. 15A–E).

Although the head and nuchal patterns in Rhadinaea decorata include all the variants seen in the combined head and neck patterns of Eutrachelophis spp., parts of the overall color pattern of Eutrachelophis are shared with other snakes as well. The overall appearance of the head and neck pattern is reminiscent, for example, of Amnesteophis melanauchen (fig. 13A, B) and Tantilla melanocephala (fig. 13C). The pattern is intraspecifically variable in the last species and probably in the first (known from a single specimen). In figure 13 an anterior band or spots (first arrow) are followed by paired spots or a fused pale crossbar (second arrow) that is medially constricted like the fused ocelli in some specimens of Eutrachelophis (compare with figs. 2 and 4).

Fig. 13.

Two unrelated small Brazilian snakes that are similar to Eutrachelophis spp. in nuchal markings and in having 15 dorsal scale rows, but they differ from each other and from Eutrachelophis spp. in fundamental hemipenial and maxillary characters. Color-pattern similarities in these and a few other small snakes include an anterior band or spots (first arrow), followed by paired spots or a fused pale crossbar (second arrow) that is medially constricted like the fused ocelli in some specimens of Eutrachelophis (compare with figs. 2 and 4). A, B. Amnesteophis melanauchen (Jan), a Brazilian snake that is known only from the type specimen collected in or at “Bahia” prior to 1863 (reproduced from Myers, 2011). C. Tantilla melanocephala (Linnaeus). AMNH R-101970 from Amazonas, Manjuru River (4° S, 57° W). Head and nuchal patterns probably are intraspecifically variable in all.

Postparietal and nuchal ocellar markings appear in Rhadinaea and other genera of “Rhadinaea-like snakes,”14 but only in a few, such as Echinanthera undulata, are they as conspicuous to the human eye as in Eutrachelophis (fig. 14A). More often, as in species of Taeniophallus, the markings represent the anterior terminus of a stripe or other body color (fig. 14B) or simply the anteriormost of a series of similar spots (fig. 15A). On the other hand, a distinct white canthal-postocular line occurs in different species of Rhadinaea and Taeniophallus but seems absent in the variational repertory of Eutrachelophis. Some of these snakes, such as Taeniophallus bilineatus (fig. 14B), share with Eutrachelophis a lateral line of whitish dashes emphasized by black edging, found on row 4 in E. bassleri (fig. 1A) and row 6 in E. steinbachi (figs. 5, 6), although these lines are nonhomologous (being edging to a dark lateral stripe in steinbachi). Similar species-specific differences in positioning of this line of pale dashes (and the black edging, which may show as a continuous narrow stripe) also occur in other genera (e.g., the decorata group of Rhadinaea).

Fig. 14.

A. Echinanthera undulata (Wied). Vivid pale nuchal spots—similar to Eutrachelophis spp.—characterize some xenodontine (and dipsadine) snakes such as this Brazilian snake (AMNH R-119764). B. Taeniophallus bilineatus (Fischer). A distinct lateral line of whitish dashes emphasized by black edging characterize Eutrachelophis spp. and other xenodontines (and dipsadines) such as this Brazilian species (AMNH R-119769); the pale canthal line seen here is variably present in “xenodontines,” but does not occur in Eutrachelophis (photographs by C.W. Myers).

Fig. 15.

A. Taeniophallus nicagus (Cope). This small Brazilian snake resembles Eutrachelophis in having 15 dorsal scale rows and an ocellar nuchal pattern. It differs in hemipenial characters from closely related congeners and from all other xenodontines (and nearly all dipsadines) in having an unforked sulcus spermaticus. B. Taeniophallus occipitalis (Jan), a Brazilian specimen. This species displays considerable variation in the head and nuchal color pattern, sometimes resembling T. nicagus above (cf., T. occipitalis in Myers, 1974: fig. 48) (photographs by Marcio Martins).

Hemipenes

Eutrachelophis bassleri has an extremely unusual hemipenis when compared with “colubrid” and “xenodontine” snakes generally. E. bassleri originally was thought to be a species of Leimadophis (now  =  Liophis), although that was ruled out once the hemipenis and its retractor muscle were seen to be unbifurcated. Nonetheless, from the appearance of the distal nude section of the retracted hemipenis (fig. 3A), it still seemed possible that it might evert as a large flattened or centrally depressed apical disc. If that were the case, it might somehow support a relationship with the Xenodontini (which have apical discs on bifurcated hemipenes). Instead, the E. bassleri organ everted with a domelike head or capitulum (fig. 3B–D).15 Although there are some similarities with the hemipenis of its unnamed sister species (fig. 4 and associated text), we have seen nothing quite like the hemipenis of E. bassleri among Neotropical snakes. A resemblance to the general physiognomy of the bassleri hemipenis is found in the Malagasy Compsophis infralineatus, as shown in figure 21; this species also has a pronounced hemipenial head or capitulum without being capitate.

On the other hand, although the deeply divided hemipenis of Eutrachelophis steinbachi (fig. 9) is markedly different from, E. bassleri, it has parallels in other taxa. Genera with similarly long-lobed hemipenes include the South American Xenodon suspectus, X. rabdocephalus, and the African, Mehelya poensis, all to be discussed and illustrated later.

Consideration of such differences might begin with the concept of a “random walk” (sensu Raup and Gould, 1966).16 The analogy between a random walk and what appears to be the phylogeny of colubrids is striking. The analogy is particularly apparent when Eutrachelophis bassleri, E. steinbachi, and other “xenodontines” are compared with the colubrids of Madagascar. One Madagascan colubrid, Mimophis, differs from other Madagascan forms, as well as from all “xenodontines,” in the greatly reduced hemipenis; it is a psammophiid and seemingly represents an independent Madagascan invasion separate from other colubrids (Cadle, 2003; Nagy et al., 2003; Kelly et al., 2008). Putting aside Mimophis, all the other Madagascan colubrids represent a monophyletic clade (Pseudoxyrhophiinae), all members of which differ from Eutrachelophis bassleri and E. steinbachi in the presence of strong hypapophyses on the posterior vertebrae, even though these two Neotropical snakes agree with all Madagascan colubrids in lack of hemipenial calyces and show a particular resemblance to Thamnosophis lateralis (but not to Liopholidophis sexlineatus) and to Dromicodryas in the construction of the orbital region of the skull. Although lack of posterior hypapophyses distinguishes many “xenodontines” from Madagascan colubrids that resemble them (e.g., the xenodontine Liophis and the Madagascar Liopholidophis), it is not absolutely diagnostic, since some “xenodontines” (e.g., Amastridium, Ninia, Nothopsis) have posterior hypapophyses, but these “xenodontines” with hypapophyses do not happen to have any counterpart in Madagascar that resembles them (the “xenodontines” mentioned have hemipenial calyces, for example). Even though the Madagascan Thamnosophis lateralis is more like Eutrachelophis steinbachi than it is like the Madagascan Liopholidophis sexlineatus in construction of the orbit, and also in the maxillary dentition with a broad diastema anterior to the enlarged last pair of teeth and in the terminal expansion of the parasphenoid, it differs from Eutrachelophis steinbachi and resembles Liopholidophis sexlineatus in a feature of the hemipenis: the organ forks only a short distance distal to the furcation of the sulcus, whereas in Eutrachelophis steinbachi the sulcus forks near the base of the organ, while the organ itself forks well distal to this, and in Eutrachelophis bassleri the unforked organ ends far distant from the sulcus furcation. But this hemipenial difference fails when we compare Eutrachelophis steinbachi with Dromicodryas (probably closely related to Thamnosophis lateralis, and with a similar orbit and with a distinctly expanded tip of the parasphenoid [expanded and also forked in the specimen illustrated by Cadle, 1996a: fig. 39, bottom]), since Dromicodryas also has the furcation of the sulcus far proximal to the terminal bilobation of the organ. However, Dromicodryas is not a “morphological intermediate,” since its maxillary dentition lacks posterior enlarged teeth and diastema (thus suggesting the Madagascan Liophidium and Micropisthodon, which it also resembles in having a reduced splenial bone, but these genera have the sulcus furcation near the furcation of the organ).

In spite of a search involving details of the skull, lungs, hemipenial morphology, and head muscles, no character has emerged that will distinguish all Madagascan colubrids from all “xenodontines.” This is precisely what would be expected if both the Madagascan radiation and the “xenodontine” radiation represent random walks with similar rules for morphological change. The same characters might easily arise here and there in both radiations, but for the same character combinations to arise in both radiations would be quite improbable.

Even without taking the comparisons to Madagascar, the question remains: How can one conceive Eutrachelophis bassleri and E. steinbachi as congeners when they have such markedly different hemipenes? We explore this question further in the Discussion below. See also the Commentary on Hemipenes as Generic and Specific Characters.

Summary of Global Comparisons

Various details have been noted in the preceding anatomical and color comparisons; few of these details need be considered important in themselves, since they involve characters known to vary among “xenodontine” snakes. However, as summarized here, some shared traits are rare or uncommon and, in total, these appear indicative of close relationship between Eutrachelophis bassleri and E. steinbachi.

Hypothesis for Hemipenial Transformation in Eutrachelophis

A few hemipenial characters seem to provide the only major structural differences between Eutrachelophis bassleri and E. steinbachi. In E. steinbachi, the hemipenis has long lobes, each lobe bearing a branch of the sulcus in a strictly centrifugal position, but with the lobe extending well past the termination of the sulcus. In E. bassleri neither the organ nor its retractor divides, although the sulcus does divide and its branches almost reach the end of the uneverted organ in a centrolineal position, becoming centrifugal upon eversion owing to differential tissue expansion. There is certainly no striking resemblance that would suggest close relationship solely on the basis of hemipenial morphology, but neither of these two species shows striking resemblance to any other “xenodontine” that would suggest an alternative phylogeny. There are slight color-pattern differences between E. bassleri and E. steinbachi, but these are common sorts of species-specific differences found in other genera and sometimes even exceeded in the variational repertory of single species (see above under Color Pattern).

Since molecular data cannot be brought to bear, the only differences conceivably of generic value are the hemipenes. The hemipenis of E. bassleri seems to be unique, that of E. steinbachi uncommon, simply because of the extent of its bilobation. In the latter, a pair of long and rather narrow organ lobes extend distally, with each long lobe having a peculiar prolongation that extends past the end of the sulcus and ends in elongated spines. The hemipenis of E. steinbachi may be concisely described as a deeply divided (long-lobed), spinose, acalyculate noncapitate hemipenis, with an apical tuff of differentiated spines and a bifurcate centrifugal sulcus spermaticus. This description also well fits the African colubrid Mehelya poensis,17 which differs mainly in size of the lobular spines (fig. 16). On the hemipenis description alone, one could claim that that Eutrachelophis steinbachi and Mehelya poensis were congeneric, but clearly the hemipenis fails at this point.

Fig. 16.

Everted hemipenis of Mehelya poensis (Smith). (AMNH R-12053, left organ.) This is a slight, deeply forked hemipenis with long slender lobes and a very long retractor muscle that originates close to the tail tip. This organ divides at subcaudal 4, with the lobes terminating at about subcaudal 21 (an indistinct join between lobe and retractor); the two slips of retractor muscle fuse at subcaudal 29, but the long muscle continues posteriorly to a broad attachment between caudals 46–49, only 10 subcaudals before the terminal caudal spine.

If growth were to stop before development of the long lobes, the organ of E. steinbachi might not be very different from that of E. bassleri. The hemipenes of Eutrachelophis spp. differ overall in appearance from hemipenes of most other “xenodontines”—lacking the apical disc of Liophis and other Xenodontini or lacking a calyculate region that is set off or not by a free edge (capitate). Both Eutrachelophis bassleri and E. steinbachi have the sulcus forked near the base of the organ, presumably indicating that the sulcus forks early in its development as it extends out onto the hemipenial rudiment.18 The proximal furcation of the sulcus and the moderately divergent (rather than nearly parallel) rami of the sulcus permit the distal ends of the rami to reach a position farther from the midline than in, for example, Hydrops marti or Rhadinaea flavilata. In Eutrachelophis steinbachi and in the everted hemipenis of E. bassleri, the rami of the sulcus reach fully lateral (centrifugal) positions on the sides of the organ; this does not simply reflect the greater length of the sulcal rami in E. steinbachi, where the sulcus branches extend onto long organ lobes missing in E. bassleri.

In Eutrachelophis bassleri the entire distal part of the simple organ is nude, but in E. steinbachi the distal lobes of the organ bear spines. However, the crotch of the organ in E. steinbachi is smooth except for small expansion folds and we suggest that this “smooth” crotch area (i.e., the smooth terminal basin) probably is homologous with the entire distal end of the organ of E. bassleri. Although the major pattern of the hemipenis develops before hatching (or birth), when the organ is everted, some details of the pattern probably arise later and may involve considerable rearrangement by differential growth. For example, mineralization of hemipenial spines to form stiff structures, rather than flexible papillae, seems to occur with sexual maturity, rather than with birth or hatching, and the definitive adult pattern of major hemipenial spines often differs from species to species, or genus to genus, in a manner suggesting differential growth superimposed on a common pattern, presumably that laid down in early development.

On the Apical Disc

It must be remembered that all that can be said of Eutrachelophis is that the hemipenial epithelium does not form projecting folds or papillae in the adult organ; just what sort of epithelial ornamentation might have been lost cannot be observed in the adult organ. Nevertheless, it is attractive to interpret the derivation of the hemipenis of Eutrachelophis from that of a Xenodontini such as Liophis for the following reasons:

Whatever the mechanics of the development of the Xenodontine apical disc may be, they would seem to be complex and unlikely to be duplicated by chance (but nonetheless duplicated by the nonhomologous apical or apicolateral discs on the hemipenial lobes of a few Asiatic and North American boids, as well illustrated in Branch, 1986: fig. 5A, B). Derivation from a stock ancestral to Xenodontini that had assembled all but one of the necessary genes for formation of discs could yield a clade very closely related to Xenodontini but lacking the taxonomic character that most clearly demonstrates this relationship.

The apical disc (fig. 17A) is a clear-cut character that nonetheless is occasionally mistaken for some other apical nude area or, perhaps more frequently, for a hemipenial lobe with an “apical depression or dimpled tip”—which usually indicates an incompletely everted lobe (but see fig. 17C caption and discussion in Myers and Cadle, 2003: 301); field eversions of some Malagasy Liopholidophis (including Thamnosophis in part) hemipenes might be thought to have apical discs except for lack of the defining, encircling lips (illustrations in Cadle, 1996a). Dowling (2004: 326) recently complained that Zaher's (1999) study of xenodontine hemipenes ignored two published descriptions of apical discs in the Middle American Conophis. For the first mentioned description, however, Myers (1986: 5: fn. 4) had already corrected it, noting:

Fig. 17.

The tips of diverse bilobed hemipenes. A. Apical discs on the hemipenis of Liophis williamsi (Roze). The apical disc is a conspicuous depressed or somewhat flattened nude area that is defined by an encircling wall, which is broached only by intrusion of the sulcus spermaticus. It resists further eversion and is a defining character of the tribe Xenodontini that has been lost in a few species (AMNH R-117671, right organ). B. An almost completely everted hemipenis of Conopsis vittatus Peters. The dimpled tips on the short lobes each conceal a small nude projection that has been confused with the apical disc (UMMZ 82650, right organ). C. Apical view showing the “umbelliform depressions” on the tips of the deeply lobed hemipenis of Thamnosophis lateralis (Duméril, Bibron, and Duméril). The umbelliform depressions resemble “dimpled tips” that usually are certain signs that the lobes are not fully everted, but in Thamnosophis the apical depressions reflect broad internal attachments of the retractor muscles and do not conceal uneverted tissue (fide Cadle, 1996a: 439, 441). (Manually everted left organ of AMNH R-60691—not completely expanded but believed to be almost completely everted with minimal dimpling).

Wellman (loc. cit.) illustrated the left hemipenis, but the right organ is still attached and shows a small dimple at the apex of one lobe and the nearly nude uneven apex on the other (fig. 17B). Although he mentioned some interspecific variation, Wellman (loc. cit.) stated, “There are no apparent hemipenial differences among the species of the genus Conophis”—a statement attributed by Dowling (loc. cit.) to a later paper by Auth et al. (1998), which was Dowling's second example of a published reference to an apical disc in Conophis.19

Clearly one must not be led astray by misinterpretation or faulty description and must remember that an apical disc is a more or less centrally flattened or inwardly sloping structure within a defining, nearly encircling wall, or lip. Cope (1894: 840) initially called the structure “a membranous disc (Disciferi),” but in 1895 he coined the term “apical disc” for use in diagnoses and keys; he clearly illustrated it with drawings of inverted organs from nine xenodontines (Cope, 1895: pls. 26 [figs. 1–Fig. 2.Fig. 3.Fig. 4.Fig. 5.6], 27 [1–2], 28 [1]). The structure has been more recently illustrated photographically by Myers (1986: fig. 7) and by Zaher (1999: figs. 81 lower, 83 lower, and 83 upper). But although seemingly reliable as a taxonomic indicator, it does not follow that absence of apical discs proves lack of relationship to Xenodontini. Unless controlled by “simple” Hox gene switches, the probable developmental complexity of these structures would mean that a considerable number of things could go wrong in their formation, leading to secondary loss of discs, as appears to have happened in several species of Xenodon, as discussed below for Xenodon suspectus. See appendix 1 for additional populations of named and unnamed species of Xenodon that have lost the apical disc.

Loss of a “Generic Character”—the Apical Disc

Hemipenis of Xenodon “rabdocephalus” (Wied):

The taxonomic value of hemipenial characters is well known, but there is limited merit to the occasionally expressed belief that they are “better” or more “reliable” than other morphological features. We call attention here to the apical disc, which is used as a novel character to define tribe Xenodontini, whose type genus is Xenodon. Myers (1986: 1) noted that the tribe had been based “essentially on a single character—the paired apical discs, which may have been lost in some populations but which are accepted as a defining synapomorphy pending further study.” See figures 17A and 18.

Myers' (1986: 1) published statement elicited a letter in 2001 from the late Garth Underwood (1919–2002). Underwood correctly assumed that Myers was referring to the hemipenis of Xenodon rabdocephalus, an exceptionally wide ranging snake that occurs from southern Mexico throughout Central America to Bolivia and northern Brazil. Underwood had been working with biometrician Clive Moncrieff and kindly summarized their work up to the year before his death:

Our holding of rabdocephalus [in the BMNH] consists of 52 specimens from southern Mexico to Bolivia and Brazil…. The most remarkable variation relates to the hemipenis; sixteen males are large enough for the retracted organ to be dissected. In all the hemipenis is symmetrically bilobed; it extends to subcaudals 12 to 20, with the cleft between the lobes from 4 to 8. Setting aside one from Honduras, the sulcus spermaticus forks at levels 2 to 4, mostly 3 (12). Spines start at level 2 or 3; the basal spines are from one to two subcaudals in length and diminish to fine spines distally on the lobes. Twelve hemipenes terminate in a bare tip; of these, five span Central America, from Belice [ =  Belize] to Panama (4 8 9 11 14), two are from Colombia (24 27) and five from Peru to Bolivia (31 33 35 37 38). One, from Manáos [ =  Manaus] (45) has lobes that terminate in a tuft of small spines. The hemipenis of the specimen from Honduras (8) extends to subcaudal ten, with the cleft at five. In the right hand organ the sulcus fork is close to the base at level one, on the left the branch sulci effectively start separately from the base of the organ! Only two specimens, 43 from Guyana and 48 from Ilhéus, Brazil, have a hemipenis that bears a terminal disc on … each lobe … This is how I come to be writing to you. I picked up that you had already noticed something strange about some Xenodon hemipenes. (letter, Underwood to C.W.M., August 19, 2001. AMNH Herpetology Archives, Myers Collection)

Regarding Underwood's reference to apical discs being present in only two (of 52) specimens—one each from Guyana and Brazil—we can add an additional specimen from Brazil. One of Wied's syntypes, herein designated lectotype has terminal discs (fig. 18). A published record (McCranie, 2011: 434) of a Honduran specimen having an “apex disc” is in error; the specimen (USNM 559716) has incompletely everted hemipenial lobes, with dimpled tips.

Fig. 18.

Hemipenis of Xenodon rabdocephalus, sensu stricto. Right retracted organ of AMNH R-3609, from Bahia, Brazil; designated lectotype herein). Abbreviations: AD, apical disc; ADW, apical disc wall or lip; Mus, dorsal and ventral slips of the major retractor muscle; SB, branch of sulcus spermaticus; SS, sulcus spermaticus below fork; STB, smooth terminal basin.

Myers quit pursuing the Xenodon rabdocephalus project after hearing from Underwood, who had managed to uncover an extraordinary amount of hemipenial variation in a named species that had appeared to maintain phenotypic similarity over an enormous geographic range. Underwood's largest sample of hemipenes (Middle America and Colombia to Bolivia), with lobes terminating in a “bare tip,” is descriptive of several additional specimens prepared by Myers. Such examples from Panama and western Colombia have spiny lobes terminating in irregular bare tips that are not disclike (e.g., fig. 19).

Fig. 19.

Everted hemipenis of Xenodon rabdocephalus, sensu lato, in sulcate view. Right organ of AMNH R-81935, manually everted (terminus of right lobe burst), from central Panama. Inset: Greatly enlarged terminus of left lobe, showing nude tip lacking an apical disc. The name Xenodon angustirostris W. Peters, 1864, is resurrected herein for populations of X. “rabdocephalus” characterized by this kind of hemipenis (compare hemipenis with apical disc in fig. 18).

Underwood also mentioned a specimen of Xenodon “rabdocephalus” from Amazonian Brazil (Manaus), with lobes terminating in tufts of small spines. That specimen may (or may not) represent a previously unnamed species, but it nonetheless is reminiscent of the following species, Xenodon suspectus, currently in the synonymy of X. “rabdocephalus.”

Hemipenis of Xenodon Suspectus Cope:

This is an uncommon snake from the upper Amazonian drainage and Andean foothills. Myers dissected the left retracted hemipenes of two Peruvian X. suspectus (AMNH R-52175, 52244) from Chanchamayo and Contamana. Both have lobes slightly longer than the base of the hemipenis, which has large spines that are replaced distally by small spines over the entire lobe except the tip, which terminates in a cluster of longer, thin spines; there is no apical disc or other nude area at the tip. The first specimen is a juvenile with nonmineralized hemipenial spines; the second specimen is an adult, whose hemipenis is shown in figure 20.

Fig. 20.

Hemipenis of Xenodon suspectus Cope; left retracted organ of AMNH R-52175 (see fig. 32). The basal half of the organ and the left lobe were opened midventrally, essentially along (and destroying) the sulcus spermaticus. Inset: Greatly enlarged terminus of left lobe, showing an apical cluster of narrow spines.

See appendix 1 for formal resurrection of Xenodon suspectus, lectotype designation for X. rabdocephalus, resurrection of X. angustirostris, and comments on the nomenclatural status of generic names based on apical disc loss.

Hemipenial Terminology for Snakes

Hemipenes are basically similar in snakes and most lizards, but enough differences have accrued among some groups of lizards to call for different sets of terms; see starting references in Savage (1997) and Ziegler and Böhme (1997). Myers and Cadle (2003: 301) provided a few references to “a recognizable hemipenial homologue” in female snakes. Eversible female genitalia are embryological rudiments and are better known in lizards, being especially well developed in varanoids. The terms hemiclitoris and the supporting structure hemibaubellum (plurals hemiclitores and hemibaubella) have been established for varanoids. More attention has been given to hemipenial and cloacal musculature of lizards than that of snakes (Arnold, 1984).

By using good illustrations, Dowling and Savage (1960) were able to standardize much of Cope's descriptive terminology for snake hemipenes. But their “medial” and “lateral” sides of the everted hemipenis have been widely replaced by “sulcate for that side of an extended hemipenis bearing the greatest length of the sulcus spermaticus, and asulcate for the opposing side” (Myers and Trueb, 1967: 236). The terms medial and lateral should be reserved for (1) the inner and outer sides of the inverted hemipenis in situ, and (2) for a pinned-flat dissected organ that has been opened along the ventral midline. Other subsequent references tweaking hemipenial terminology for snakes include Cadle (2010, 2012a), McDowell (1961), Myers and Campbell (1981), Branch (1986), Savage (1997, 2002), and Zaher (1999).

The orientation of the sulcus spermaticus is of taxonomic importance. In the case of simple (single) sulci, much evolution has involved loss of one branch of a forked sulcus—normally the left branch in colubrines, the right branch in natricines. The descriptors sinistral sulcus and dextral sulcus are useful for designating simple sulci extending, respectively, either to the left or to the right hemipenial lobe (Rossman and Eberle, 1977: 40). The term centrolineal sulcus applies to an unforked sulcus that appreciably declines neither to the left nor to the right. Savage (2002: 539) coined the term semicentripetal for a simple sulcus in the center of a simple (unlobed) hemipenis when the right and left hemipenes are mirror images; Cadle (2012a: 219–220) thought that “semicentripetal” is likely to be confused with “centripetal hemipenes” (most common among natricines) and that “centrolineal” better serves the situation described by Savage, even though that name originally was proposed by Myers and Campbell (1981) for use with forked sulci on bilobed organs (see Cadle and Savage, 2012: 36).

“The ‘left’ and ‘right’ hemipenial lobes and/or sulcus branches refer to the everted hemipenis as viewed looking toward its sulcate side (hemipenial lobes are dorsal and ventral when retracted)” (Myers, 2011: 14, fn. 6). Whether the paired hemipenes are symmetrical or asymmetrical should be included in descriptions more often: Underwood (1967: 45–46) thought that asymmetry defined his family Colubridae and that symmetry his family Dipsadidae (“with the exception of the Sibynophinae”). A common situation in either group is for the sulcus spermaticus to enter the hemipenial base in mirror image but to become symmetrical or nearly so distally (e.g., showing the sulcus entering from the left on the left organ and from the right on the right organ: Dowling and Savage, 1960: fig 1; this paper fig. 25A, B).

McDowell (1961) and Myers and Campbell (1981) coined a set of three terms to describe the orientation of forked sulci spermatici in either bilobed or single hemipenes—centripetal, centrolineal, and centrifugal. That usage (at least for the first two terms) has been extended also to cover simple or unforked sulci. Most workers seem to have found these terms applicable for a variety of snakes (e.g., McDowell, 1987, in discussion of colubroid subfamilies). McDowell (1968: 570) had earlier suggested the terms revolute-centripetal and orthro-centripetal for a few configurations. Branch (1986: 287) believed that “a fuller standardized terminology is necessary” and elaborated on McDowell's approach with the following eight descriptors (having “sulcus” hyphenated to each): semicentrifugal, centrifugal, semirevolute, revolute-centrifugal, semicentripetal, centripetal, semirevolute-centripetal, and revolute-centripetal. The system, however, can be hard to remember and lacks generality for snakes globally. To paraphrase Savage (1997: 23), there appears no need to coin any special terms for features that may be simply described.

We suggest keeping the simple tripartite system summarized in Myers and Campbell (1981: 16–17) by extending coverage to simple (unforked) sulci spermatici and by appending just a few extreme conditions for use only when needed, thus without sacrificing generality (i.e., usefulness):

Fig. 21.

Distinct hemipenial capitula (heads) with and without capitation. A. Rhadinaea taeniata (Peters). Retracted and everted views showing a capitate capitulum set off by a well-developed overhang that is interrupted only by entrance of the sulcus spermaticus (UMMZ 121522, left retracted and right everted organs; from Myers, 1974: 107, fig. 23A, B). B. Compsophis infralineatus (Günther). Sulcate and three-quarters views of a well-developed capitulum lacking capitation (MZC 181153, everted; from Cadle, 1996b: 75, fig. 17, as Geodipsas). The capitula are ornamented with calyces in A and with small spines in B. (See also fig. 3 for a distinct noncapitate capitulum that is distally nude and proximally spiculate.)

Sulci spermatici are not constrained by such classification. Unusual or unique variations can be simply described with as little jargon as possible.

Attention also needs to be given to single descriptors that are in use and that need some clarification in our opinion. The terms capitulum and capitate have a common root and could be (and have been) treated as synonymous. But a capitulum (plural capitula) is simply a “head”—a visually differentiated part of the distal end of the hemipenis (e.g., as in Eutrachelophis bassleri, fig. 3). The terms capitate and capitation (literally “having a head” or “headed”) were redefined or restricted by Myers (1974: 31) in order to refer to the common condition where a capitulum is set off by an overhang. This reflects Cope's (1895: 201) usage—“capitate (or pocketed at base of calculate portion)”—and is the only meaning of “capitate” in Dowling and Savage (1960: 25, fig. 6B). But ornamentation of well-defined capitula is not confined to those that are capitate, as shown in figures 3 and 21.

If bilobed organs have separate (noncontinuous) capitula, they are bicapitate if each capitulum is separately set off by an overhang. Although this “overhang” or “edge” is not a groove like a sulcus, Dowling and Savage (1960: 26) used the term “deep groove” and Zaher (1999: 9) coined “capitular groove” for Myers' “free overhanging edge except where crossed by the sulcus spermaticus” (Myers, 1974: 31). The term semicapitate was used by Myers (1973: 7–8) for retracted hemipenes of Saphenophis, in which the calyculate area was predicted to retain a slight overhang on eversion (as shown by Zaher, 1999, fig. 76). “Semicapitate” is now commonly used for describing a weak or interrupted overhanging edge of a capitulum. Zaher (1999: 8–12, fig. 1) combined and illustrated many hemipenial terms involving degree of capitation and calyculation, distribution of spines, etc., which allowed him to summarize and contrast generic characteristics.

Dowling and Savage (1960: 26) mentioned that some “Atractus have what appears to be a third type of basal ornamentation, a basal naked pocket. However, no fully everted hemipenes … are available, and the true appearance of this structure is not known.” The basal naked pocket is now known to occur widely in everted organs of some Atractus and other genera and “often is conspicuous even when the hemipenis is fully everted” (Myers, 1974: 32). Various distal naked pockets are associated with the edges of the capitular overhang, as in some Imantodes (Myers, 1982: figs. 4, 18, 19), and Pliocercus and Urotheca (Myers, 1974: 273, fig. 4).

An ignored, possibly primitive character needing attention is the smooth terminal basin in the crotch region of some bilobed hemipenes; it is shown as STB in figures 9 (drawing) and 18 (photograph). Although the crotch is smooth in many colubroids, in the case of the STB it is well defined by an abrupt change to spinules or calyces at its rim. It is like the sulcus in being smooth, but it is composed of much thicker tissue. The sulcuslike smoothness can be seen in figure 18; direct examination of the inverted organs of Eutrachelophis steinbachi reveal what appear to be expansion folds in the STB (as shown in fig. 9), suggesting that the basin is relatively large in the everted hemipenis (fn. 7). In many species the sulcus spermaticus forks and runs centrolinearly or centrifugally to bypass the terminal basin or meet it at its most distal ends.

Other terms for the ornamentation and armature (spines) of the everted hemipenis are well covered in Dowling and Savage (1960) and are transferable to the anatomy of retracted organs. Without complementary comparisons of retracted and everted organs, it may be impossible to imagine hemipenial shape acquired during either eversion or retraction (e.g., compare fig. 25C, D). Differential tissue expansion during eversion sometimes leads to unimaginably large and odd shapes, but on retraction there has to be a method of folding and storing the tissue. Retraction may provide a few additional characters, such as whether the asulcate side of the capitulum makes a single or a double fold, a usually interspecific character but sometimes intraspecifically variable (Myers, 1974: table 2 and various figs.).

The ontogenetic stiffening of spines and/or spinules is an indication of sexual maturity, as recognized by Cope's (1895: 189) assertion that “spines are not ossified in young snakes.” Following Cope, Myers (1974: 31et passim) incorrectly used “ossified” and “unossified,” but bony structures in squamate hemipenes seem to occur mainly in varanids (Shea and Reddacliff, 1986; Ziegler and Böhme, 1997). The term calcification is commonly (and probably correctly) used for the stiffening of spines and spinules, but mineralization perhaps is better in not presupposing the nature of the hardening.

The retracted organ rarely needs new terms, but Myers and Cadle (1994: 12–13) coined pseudocalyculation and pseudocalyces for calyxlike structures that disappear on eversion. Zaher (1999: 11) viewed pseudocalyces as “reduced calyces”; Dowling (2004: 321) flatly asserted that “the disappearing calyces” were simply “an artifact of incomplete specimen preparation” (i.e., apparently by not injecting or somehow filling the “blood sinus (or the central cavity) … and then the outer lymph sinus” [emphasis added]).22 In response to Dowling, Myers (2011: 16, fn. 11) provided further description of pseudocalyces and carefully explained that “Normal calyces have walls of varying flexibility but do not disappear no matter how the organ is everted or to what extent it is inflated.” Myers and Cadle (1994: 13) stated that they could not “disprove possibilities that [pseudocalyces] are either primitive or vestigial calyces, [but suspected] that they simply represent a method of folding expansible tissue when the … hemipenis is inverted for storage.” In any case, pseudocalyces have since been detected in: (1) Dendrophidion brunneum, with a subtle difference that the transverse walls are retained as circumferential flounces in the everted condition (Cadle, 2010: 19–20), and (2) Phyllorhynchus browni (Cadle: 2011: 6). And there may be many taxa with very enlarged and/or weakly defined calyx-shaped structures that cannot be confused with “normal calyces” on either retracted or everted organs (e.g., see Cadle, 2010: 19–20). The term “pseudocalyces” proved to have first been used by Broadley (1980: 492, 506, pl. IIA) for the quite different, peculiar anastomosing hemipenial flounces in two subspecies of Prosymna sundevallii (Broadley's illustration reproduced herein as fig. 29C).

Finally, now that techniques are available for everting hemipenes of preserved snakes, it must be remembered that such “manual eversions” often are less rotund when compared with “field eversions” obtained from freshly killed specimens. The organ is fully everted if all surfaces and structures are visible, including the terminus or termini of the sulcus spermaticus. It is maximally expanded if it is fully everted and inflated to the fullest extent allowed by its original elasticity (discussion in Myers and Cadle, 2003: 295).

Are Hemipenes “Conservative” Characters?

The question whether the hemipenis is a conservative character poses a conundrum. Answers probably will depend on the taxonomic groups with which one has had experience. Bogert (1940) and Arnold (1986) are among the few herpetologists who have wrestled with the problem specifically, for snakes and lizards respectively.

Many recent (younger) workers seem to have accepted Dowling and Savage's (1960: 19) assertion that “It has been amply demonstrated that the features of the [snake] hemipenis are rather stable for a particular species or species group.” Since then, it seems commonplace to assume that examination of one or a few hemipenes is sufficient to typify a species or a genus. Dowling (1967: 138) later followed up with the statement that the hemipenis “has no obvious correlation with the ecology, food habits, or locomotion of the animal. Such an uncorrelated structure may give better information on genetic relationships than some habit or habitat-correlated –characteristic.” This is in marked contrast to conclusions reached by Bogert (1940: 9–10) after his survey of hundreds of hemipenes (and other characters) in African snakes; Bogert noted Noble's belief that the courtship pattern of snakes “has not undergone a great change in phylogeny”23 and then went on:

Bogert's descriptions had to be based almost entirely on retracted hemipenes since techniques of manual eversion had yet to be developed. Comparisons of Bogert's descriptions of retracted hemipenes with organs actually everted reinforce the reality that hemipenes are evolutionarily dynamic. Generalities implying that genitalia are relatively free of selection pressure are wrong. In addition to Bogert's observations correlating hemipenes with habits and habitat, there is a growing body of evidence pointing to the existence of powerful sexual selection on male genitalia over a wide range of organisms. It will be noteworthy to show whether there are any closely related species of snakes that can not be differentiated by hemipenial characters (not the same thing as pointing to related species in which the hemipenes seem almost identical). At least for species with large distributions, intraspecific variation—part of the stuff of evolution—seems to us to be the rule, as suggested by the following examples where investigators have bothered to look at more than a few specimens; examples are given in chronological order.

Examples of Intraspecific Variation in Snake Hemipenes and Cloacae

Some Basics:

Browsing through chapters in a recent book on reproduction in snakes (Aldridge and Sever, 2011), one finds the following sentence: “Each hemipenis is equipped with two lobes (bifid), each with a sperm transferring branch (the sulcus spermaticus) which fits exactly into the female cloaca, the spines on the hemipenis holding it in the correct position for transferring sperm into the two oviducts in the bilobated cloaca.” That simple declaration was derived from an important paper by Pope (1941). But, in that paper, Pope himself had commented: “One unfamiliar with snake genitalia might conclude that [my] observation … virtually closes the problem of hemipenial adjustment, whereas actually the problem is barely opened.”

Indeed, 70 years later we really are not much further along. Accomodation to the hemipenis by the female cloaca or vice versa remains a subject virtually untouched. When viewing the diversity of hemipenes illustrated in this paper and elsewhere, one should wonder: How does it fit? And in what way, if any, is the cloaca modified? Ultimately such questions lead to the topic of sexual selection and “sperm competition,” with growing awareness that there are degrees of difference between snake and lizard hemipenes, both in anatomy and functional use (see the extended treatment by Olsson and Madsen, 1998; also referenced below under Conclusions and Brief Summation).

Pope (1941) illustrated the hemipenis and the cloaca of two snakes (Liophis poecilogyrus) “killed while copulating and preserved without dislodging the penis.” His illustration is instructive and reproduced herein as figure 22. Pope probably did not realize that the shallowly bilobed Liophis hemipenes terminate in apical discs (as in fig. 17A), which, however, fit his description that: “Dissection shows that each branch of the sulcus ends in a lip surrounded by a ridge which is firmly pressed against an area of the cloaca, in the center of which an oviduct opens.” The “ridge” labeled in figure 22 is the outer part of the apical disc that is broached by the inwardly sloping sulcus spermaticus. The apical disc thus makes a snug fit against the opening to the oviduct and helps pool seminal fluid. Pope (op. cit.: 250) drew attention to Cope's (1900) comments on the comparative anatomy of the cloacal region:

Fig. 22.

Illustration (from Pope, 1941: 251), showing copulatory adjustment between hemipenis (left) and female cloaca (right) of Liophis poecilogyrus, as seen by dissection after the snakes had been killed while mating. Pope commented: “The ventral body wall of the female has been bisected and its halves spread apart to expose the interior of the cloaca, from which the turgid hemipenis was removed. The exposed side of the latter lay against the dorsal wall of the cloaca so that each sulcus spermaticus opened at the orifice of an oviduct.” Species of Liophis have an apical disc on each hemipenial lobe (as in fig. 17A), which should confine the flow of seminal fluid to the opening of the oviduct; in this case there may have been a slight postmortem retraction of the discs. The area labeled “ridge” in this figure corresponds to the apical disc itself, which contains the terminus of the sulcus spermaticus.

a common chamber or vagina, which is above the rectum, and opens into the cloaca. This vaginal chamber is large, and is divided more or less completely in the Solenoglyph snakes, is about half divided or deeply bilobate in the Colubroidea, and is undivided in the Peropoda. Its external wall is deeply longitudinally grooved, and the internal wall is transversely grooved in Crotalus. In Colubroidea generally it is deeply longitudinally grooved on all sides. In the Peropoda it is nearly or quite smooth. The cephalic extremity of the oviduct is for a short distance transversely plicate or lobate, the labia being held in place by simple unfolded bands of the inferior and superior edges. The fontanelle is immediately cephalad of this region, and has very thin simple walls. Being only a slit, it is sometimes difficult to discover. The oviducts do not accompany the ureters so closely as do the vasa deferentia, and approach nearer the middle line for a short distance below the rectum. (Cope, 1900: 700)

Pope (op. cit.: 250–251) gave his own observations on matching hemipenial and cloacal fit in two species of Asiatic Trimeresurus in which females could not be distinguished by external characters alone.

Edgren (1953) published a second article on copulatory adjustment, based on two specimens of Heterodon platyrhinos that were killed and preserved in coitus. However, Edgren (op. cit.; 163–164) failed to find the intimate relationship between hemipenis and cloaca reported by Pope; the hemipenis partially everted “through the area of distal large spines and only slightly into the calyculated area. The terminations of the sulcus branches … were near the openings of the oviducts, but there was no close application [and the hemipenis] filled only about three-fourths of the cloaca … and may represent a post-mortem artifact.” The method of killing was not mentioned, but it does seem likely that the hemipenis had started to relax and retract.

Even the fact that only a single hemipenis is used in a copulatory event did not become established until the 20th century (see Pope, 1941: 249). An erroneous report of simultaneous use of both hemipenes (Brain, 1959: 71, fig. 1) was based on a mating by Pseudaspis cana, which has extraordinarily long hemipenial lobes that are joined close to the base, as can be seen in figure 28. Zweifel (1997) compiled data for Lampropeltis suggesting that hemipenes are alternated if a mating occurred within several days, but that choice probably is random after “longer periods of [enforced] abstinence.”

Variation in Hemipenis and Cloaca of Calamaria lumbricoidea:

Shortly after the appearance of Dowling and Savage's survey, Inger and Marx (1962: 37) warned that “individual and geographic variation should be analyzed before undue taxonomic weight is given to the hemipenis and cloaca.” They described remarkable variation in the hemipenis and cloaca of the widespread Southeast Asian snake Calamaria lumbricoidea H. Boie. They found three types of hemipenes (simple, smooth; forked, smooth; and forked, papillate) and three types of cloaca (bulbous, cardioid, and bilobed). But they found no correlation between the type of hemipenis and form of the cloaca, which “appears to be constant within populations, but variable between populations.” This was a rare study to consider cloacal shape and their illustration is reproduced here as figure 23. In their follow-up generic monograph, Inger and Marx (1965: 85) said the “hemipenis of lumbricoidea shows variation not seen in other species of Calamaria.”

Fig. 23.

Illustration (after Inger and Marx, 1962: 36) showing intraspecific variation in the female cloaca of Calamaria lumbricoidea. Diagrammatic ventral views, with intestine cut and reflected. A. Bulbous cloaca. B. Cardioid type. C. Bilobed type. D. Bilobed type with ventral wall removed.

Hemipenial Variation in the Vine Snake Oxybelis aeneus:

Keiser (1969: 117; 1974: 16–17, fig. 1) examined the ornamentation of one or both hemipenes for about 60% of the available males and determined origins of the paired major retractor muscles for 68 specimens.24 Keiser (1974: 16) noted considerable variation “in the number, size, and arrangement of the basal spines, the caudal level at which the spines occurred, the size of the hemipenes, the arrangement of the denticulated calyces, the nature of the apex, and the caudal level at which the retractor penis magnus muscles originate.” His figure 3 shows photographs of two very different looking hemipenes on Mexican specimens; one is robust with well-developed basal spines, the other is slender with reduced basal spines—these variants are geographically correlated, but the nature of the interpopulational variation “necessitates a degree of subjectivity when comparing them” (Keiser, 1974: 16–17). Keiser (loc. cit.) described the sulcus spermaticus as “deep, single, and situated laterally on the inverted hemipenis. It occasionally forks near the apex.” The last statement, for a long time overlooked, was in fact surprising, inasmuch as Oxybelis has always seemed firmly ensconced in the Colubrinae. More recently, however, forking, including terminal forking and/or terminally divergent lips of the sulcus have been documented for some other Neotropical colubrines (Dendrophidion spp.: Cadle, 2010, 2012a, 2012b; Freire, et al., 2010); forking also occurs in some Asiatic Ahaetulla (McDowell, 1987: 41), so a simple, unforked sulcus might seem not to be as universal a colubrine trait as has been supposed (e.g., Dowling and Duellman, “1974–1978” [1978]: 112c.1; Underwood, 1967: 46, 149). In any case, the Oxybelis hemipenis could use further attention, particularly concerning the nature of the small type of hemipenis as well as variation in the sulcus spermaticus.

Keiser (1974: 41) concluded that “Variation in hemipenial size and ornamentation is great in all parts of the range, but most striking in individuals from southern Zacatecas, Sinaloa, and the Tres Marías Islands.” Despite questions raised, Keiser clearly found significant inter- and intrapopulational variation in this Neotropical vine snake.

Examples in “Rhadinaea”:

Myers (1974: 30) thought that

the hemipenis frequently seems to provide proof of relationship when divergence and convergence have obscured the usefulness of scutellation, color pattern, and dentition … [but] this is not to claim that taxonomic use of the hemipenis is not liable to the same pitfalls that beset other characters. Evolution does not permit universally reliable taxonomic criteria, and even the snake hemipenis is not so conservative as formerly thought.

Myers surveyed hemipenial structure in “Rhadinaea” and some Rhadinaea-like snakes (supra, fn. 14), using both retracted and everted hemipenes when possible. The eight resulting species groups appear to be mostly (if not entirely) monophyletic. The groupings received hemipenial support as summarized in Myers' tables 2 and 3 (Myers, 1974: 34f., 42f.), but species differences were also pointed out and become more obvious when viewing some of his figures (especially figs. 9, 20, 30).25 Two intriguing examples of variation in Rhadinaea can be mentioned here: (1) The retracted hemipenes of Rhadinaea vermiculaticeps and “R.” schistosa seem virtually identical even to the unusual straight spines; the former is the generic type species; the latter, formerly in the Rhadinaea godmani group, is the type species of Rhadinella. The correspondence between their hemipenes can be seen by comparison of figures 30G and 39 of Myers (1974: 133, 165). When material becomes available, it will be interesting to compare the everted organs. (2) The snake Rhadinaea decorata has a large distribution (Mexico to Ecuador) and considerable variation in head and nuchal color pattern that exceeds the variation in genus Eutrachelophis of this paper. And the hemipenis of R. decorata varies extraordinarily in length, so much so that that some southern specimens seem to call out for status as a separate species (see fig. 24). However, nomenclatural action was not taken because of the nature of geographic variation in the length of retracted organs, which span the lengths of 5 to 8 subcaudals in Mexico, and 6 to 14 subcaudals in Costa Rica and Panama (Myers, 1974: 75). DNA evidence and more everted organs may someday provide insight.

Fig. 24.

Inter- and intrapopulational hemipenial variation in Rhadinaea decorata (Günther). A. Short bulbous hemipenis, 7 subcaudals long, Veracruz, Mexico (TCWC 21391, right organ). B. Long slender hemipenis, 11 subcaudals long, Costa Rica (USNM 120832, right organ). Geographic variation is complex, comprising lengths of 5–8 subcaudals in Mexico and 6–14 subcaudals in Costa Rica through Panama. Field eversions, drawn to same scale, from Myers, 1974: 74–75.

Pythoninae, Boinae, and Acrochordoidea in New Guinea and the Solomons:

McDowell (1975, 1979) included these groups in his catalogs of noncolubroid snakes in New Guinea and the Solomons and summarized intraspecific hemipenial variation when there were sufficient specimens. In a sample of 30 Python amethistinus there was variation in length of hemipenis (7–10 subcaudals), point of forking, extent of calyculation, and number of flounces (3–6). A sample of 19 Chrondropython viridis varied in length of hemipenis (5–13), whether feebly bilobated or single, and extent of calyculation. The hemipenis in species of Liasis varies from bilobate to deeply forked; the tip of the sulcus on each lobe is “prolonged on a smooth papilla that projects abruptly and in nipple-like or loop-like fashion from the untapered lobe” (McDowell, 1975: 31); in some specimens of L. albertisii the retractor muscle is attached not to the extreme tip of the papilla but to the base, so that the papilla remains everted within the inverted organ. There is variation in length of the hemipenis at least in L. papuanus (8–12 subcaudals).

Among the Boinae, the hemipenis of Candoia bibroni varies in length from 9 to 20 subcaudals, but there are no significant differences between islands and the two extremes come from the same island (Santa Cruz). C. carinata and C. aspera also vary in hemipenial length (7–18 and 5–11 subcaudals, respectively), with significant differences between some localities for the former but not the latter. There are substantial differences in the hemipenes of the three species of Acrochordus, as treated under below under Evolutionary Plasticity and Extreme Divergence.

Variation in North American Snakes Related to Tantilla planiceps:

Cole and Hardy (1981) examined inter- and intraspecific variation in eight species They found intraspecific variation in aspects of calyculation and in the distribution and numbers of spines and spinules, but each was found to have a species-specific combination of traits that characterize its hemipenis. The physiognomy of all but two hemipenes was each quite different, varying in number of spines, nature of the apical calyces, and presence or absence of capitation (see their fig. 6). The two most similar species (T. atriceps and T. nigriceps) were usually separable by the place of origin of the major retractor muscle. The origin of this muscle varies intraspecifically in T. hobartsmithi, the externally most distinctive species with a capitate capitulum. Cole and Hardy (1981: 203) summarized that “The most useful characters for distinguishing species of Tantilla, particularly in North America, appear to be in the hemipenes and head coloration.” That generalization seems likely to hold for most of this large genus.

Loss of the Bifurcated Sulcus Spermaticus in Tribe Imantodini:

As cited above, Bogert (1940: 9) believed that an unforked sulcus spermaticus had evolved numerous times in Old World snakes. But the only New World “xenodontines” known to have an undivided sulcus spermaticus are several species of Dipsadinae (i.e., Imantodes spp. and a few Leptodeira spp., the sole members of tribe Imantodini; see Myers, 2011: 21–26).26 Imantodes and some Leptodeira are defined by an unforked sulcus spermaticus, whereas other Leptodeira may retain a tiny terminal fork or have the sulcus distally expanded without forking—i.e., both inter- and intraspecific variation are evident. In that analysis, Myers (2011: 23) concluded that there generally appears “to be at least four ways in which primitively sulci spermatici can become single”: (1) by simultaneous shortening of both branches; (2) by shortening of a single branch; (3) by shortening and/or loss of a single hemipenial lobe; and (4) by weakening and loss of the medial sulcus lips, leaving only a tissue divide separating the lateral branches. The first and fourth mechanisms seemed to apply to Leptodeira spp., whereas the second method fits the evidence for Taeniophallus immediately below.

Loss of the Bifurcated Sulcus Spermaticus in Taeniophallus:

In the New World, a forked sulcus spermaticus has long been recognized as a defining character of the Xenodontinae, sensu stricto, including at times a few other “xenodontines” (i.e., Dipsadinae, tribe Imantodini, see above). However, an exception for Xenodontinae has been confirmed. Despite the “remarkable similarity” of Lygophis nicagus to “Rhadinaea” brevirostris, Myers (1974: 208) had concluded they were not congeneric and that nicagus probably was not a South American snake. Myers and Cadle (1994: 5–6), however, proved the South American provenance of nicagus. Although Myers (1974: 32) had commented that shortening one branch of the sulcus spermaticus could lead to a simple, unforked sulcus, he had retained enough faith in hemipenial characters not to make the speculative leap required for placing nicagus in the same genus with brevirostris.

As now recognized Taeniophallus nicagus and T. brevirostris appear to be sister species that demonstrate evolutionary stages from a bilobed hemipenis with a forked sulcus to a unilobed hemipenis with a simple sulcus. See figure 25.

Fig. 25.

Evolutionary loss of a bilobed hemipenis and a forked sulcus spermaticus, demonstrated by apparent sister species. A. Taeniophallus brevirostris (W. Peters) showing retention of weak bilobation (right lobe incompletely everted) and shortening of the left branch of the sulcus spermaticus (MZUSP 8484, left organ). B. T. brevirostris showing complete loss of bilobation and similar shortening of the left branch of the sulcus spermaticus (AMNH R-28799, right organ). C. Taeniophallus nicagus (Cope) showing complete loss of bilobation and complete loss of bifurcation in the now-single sulcus spermaticus, which extends from the base to one side of the apex (AMNH R-138683, right organ). D. T. nicagus, an uneverted hemipenis, showing the unbifurcated sulcus spermaticus that extends to the apex, where it is largely concealed by the close-pressed sulcus lips (MCZ 149545, left organ). The first three hemipenes were manually and fully everted (except for one resistant lobe in A), but they probably are not completely expanded as is commonplace with eversions made from preserved specimens. Not to same scale.

Intraspecific Variation in Hemipenial Eversion:

Zaher and Prudente (1999) found that species of Siphlophis and the related Tripanurgos compressus evert their bilobed hemipenes in both Y-shaped and T-shaped conditions. Dowling (2002) asserted that the apparent polymorphism was simply an artifact of preservation, caused by tension of the major retractor muscle (which can indeed sometimes affect the shape of a preserved hemipenis [e.g., Myers and Cadle, 2003: fig. 1]). However, Zahr and Prudente (2003) satisfactorily answered Dowling by further demonstrating that the eversion polymorphism is real, not artifactual.

Unusual Hemipenial Morphology in Phyllorhynchus:

Cadle (2011: 4–6) described an unusual morphology in hemipenes of North American Phyllorhynchus, in which the sulcus is not divided as previously reported, but in which the sulcus lips distally diverge and open onto an extensive apical nude region. He also reported intrageneric variation in which calyces were either reduced or absent. He described and illustrated presumed intraspecific variation for P. browni, which has two distinct hemipenial morphs, differing in presence or absence of bilobation and nature of the spines.

Inter- and Intraspecific Variation in Dendrophidion:

Cadle (2012a) discovered cryptic species of the Middle American Dendrophidion vinitor complex based on hemipenial characters. He expected, but did not find, something similar in his redefined Dendrophidion percarinatum, which ranges from Honduras to Colombia. As he concluded for D. percarinatum, “there is variation in virtually all aspects of hemipenial morphology, including the form of the spines and the degree to which fully formed apical calyces are developed” (Cadle: 2012b: 319).

Evolutionary Plasticity and Extreme Divergence of Snake Hemipenes

Eutrachelophis and Acrochordus:

Despite contradictory hemipenial evidence, global comparisons of anatomy and color pattern show that Eutrachelophis bassleri and E. steinbachi are so closely related as to be considered congenerous in our opinion. It is certainly a case of extreme divergence in hemipenes, as seen by comparing figures 3 and 9. As stated, our simple explanatory hypothesis involves embryonic suppression of the hemipenial lobes in the bassleri lineage and homology between the hemispherical nude apex of E. bassleri and the interlobular smooth terminal basin in E. steinbachi.

A similar scenario might also be applicable to major hemipenial differences between Acrochordus arafurae, A. javanicus, and A. granulatus, as set forth by McDowell (1979: 69, fig. 26). All three species have a moderately elongated hemipenis, but A. arafurae differs conspicuously from the other two in that: (1) the organ is bilobated only at the extreme tip, rather than being forked for a third to a half of its length; (2) the sulcus is forked near the tip of the organ, rather than near the middle; (3) there are no spines, papillae, calyces, or flounces, in contrast to the other two species in which the sulcus is forked more proximally and there are numerous spines or papillae (probably depending on age) on the lobes excepting the distal ends, with a few spinules extending more proximally along the sides of the sulcus on the unforked portions of the organ.

Fig. 26.

Two small Panamanian snakes, genus Enulius— anatomically similar and apparently closely related—but recently separated generically on the basis of what appears to be extreme variation in the hemipenes (see text and fig. 27). A. Enulius flavitorques (Cope), with a yellow nape band (AMNH R-109600, Canal Zone, Ancon). B. “Enuliophis” sclateri (Boulenger), with an enamel white head (KU 112619, San Blas, Camp Sasardí. Both photographed larger than life by C.W. Myers).

Acrochordus arafurae might differ from the other species in a single developmental character—a delay in the forking of the sulcus until the hemipenis has nearly reached its limit of length—based on a few simple assumptions: (1) there is a limit imposed (perhaps by length of the retractor muscle) on how long the entire hemipenis can be; (2) furcation of the organ cannot take place until after the sulcus has forked; and (3) spines begin formation only proximal to a nude apex on a distinct hemipenial lobe, when the very short lobes of arafurae would correspond to only the nude tips of the lobes of javanicus and granulatus and a threshold for initiating spine formation had not been reached. Thus, differences between the three living species of Acrochordus could all be expressed as quantitative differences in the timing of furcation of the sulcus (earliest in A. javanicus, latest in A. arafurae, with A. granulatus intermediate) relative to the time when the sulcus reaches the end of the organ. Descriptively, A. arafurae has lost hemipenial spines and those who construct phylogenetic hypotheses from descriptive characters would probably regard it as impossible for A. arafurae to give rise to a species having spinose hemipenial lobes. But considered from a developmental scenario, it is not known whether A. arafurae has hemipenial spines or not; the area where spines would be expected is not formed and we are as ignorant of this particular feature of the hemipenis as we would be if all available specimens were females. It is conceivable that if a species derived from A. arafurae were to speed up sulcus furcation, long hemipenial lobes with spines would reappear. Considering developmental scenarios rather than descriptive characters seems to introduce doubt where previously there was certainty, but if the certainty was an illusion, then this is all for the good.

Enulius and “Enuliophis”:

Enulius flavitorques (Cope) and Enulius sclateri (Boulenger) share very distinctively modified maxillae and seem to differ only slightly in scutellation and color pattern (fig. 26). The close similarities were pointed out by Dunn (1938: 417), who first recognized the two species as congeneric and applied the available name Enulius; Dunn described the maxillary dentition as “3–4 small teeth increasing posteriorly, followed immediately by one or two relatively enormous flattened (ungrooved) fangs…. No American snake is known with … similar dentition.” See McCranie and Villa (1993: fig. 2) for a drawing of the dentition, which is reminiscent of the Malagasy Exallodontophis albignaci, except that the latter has two large teeth intervening between the anterior small teeth and the posterior large fangs (see Cadle, 1999: fig. 10). Dunn (op. cit.) did not know the hemipenis of Enulius sclateri, but described that of E. flavitorques as “slightly bifurcate, sulcus forked at extreme distal end, no calyces, organ with minute uniform spines.” This is accurate except for the presence of distinct apical projections or awns, shown for the first time in figure 27A, B.

Fig. 27.

A, B. Hemipenis of Enulius flavitorques in sulcate and asulcate view. Left organ of AMNH R-107406, measuring 6.5 mm from base to crown between apical projections, which are about 2.5 mm long. C. Hemipenis of “Enuliophis” sclateri, roughly ×6.3. From McCranie and Villa (1993: fig. 1, “organ 10.6 mm”).

McCranie and Villa (1993) found that the hemipenis of Enulius sclateri has a scattering of large spines in additional to small spines that are not uniformly distributed. Their illustration is reproduced here as figure 27C; it looks as though the bilobed hemipenis is completely everted, without evidence of terminal projections. In following prevailing views on the taxonomic importance of the hemipenis, McCranie and Villa (1993: 262) thought that the “distinctiveness of the sclateri hemipenis warrants the designation of a new genus to accommodate the species.” In not recognizing “Enuliophis” as a valid genus based solely on the hemipenis, we agree with Savage (2002: 590), who succinctly stated: “In all other features E. sclateri clusters with Enulius when compared with other colubrid snakes. Since all evidence indicates that E. sclateri is the sister species to all other Enulius, taxonomic efficiency is best served by avoiding monotypic genera and including the known species in an inclusive taxon.”

Coiled and Folded Retractor muscles and a Folding Hemipenis:

If the maxim “size matters” is sometimes true, how does a snake balance selection for a long hemipenis against selection for a short tail, since both sometimes seem to be advantageous? It is been done differently in clearly unrelated snakes, in which hemipenes are nearly as long or even longer than the tails that must house them.

In Scolecophidia, McDowell (1974: 6–7) summarized ways in which maximum-length hemipenes are housed in short tails of species from New Guinea and adjacent regions:

The hemipenis of Typhlops has the long retractor muscle inserted at the extreme tip of the relatively short and broad organ; hence the entire organ is inverted when the organ is retracted. In Typhlina, the long retractor is inserted at about the middle of the organ and the distal part of the organ remains permanently everted and is an essentially solid awn bearing the distal part of the sulcus spermaticus; the proximal, inversible part of the organ acts as a sheath for the permanently everted portion when the hemipenis is retracted. Usually the hemipenis and its retractor are straight in Typhlops, but in the single T. inornatus examined … the retractor has a zigzag bend. Usually the hemipenis and its long retractor show helical coiling in Typhlina (at least 2 coils in all New Guinea and Solomons species), but in the [long-tailed] Philippine T. cumingii … the organ and its retractor are either perfectly straight or have a single Z-shaped flexure in a single plane.

Some Alethinophidia are unusual in having the major retractor muscle originate close to or even at the tail tip, which correlates with a relatively longer hemipenis. The African colubrid Mehelya poensis, for example, is not a particularly short-tailed snake, but its deeply divided, long-lobed hemipenis (fig. 16) extends to about the level of subcaudal 21 and the long retractor has a broad anchorage between subcaudals 46–49, only 10 subcaudals before the terminal spine.

Another colubrid, Pseudaspis cana, has an impressively large, more deeply divided and longer lobed hemipenis, as seen in figure 28. In the specimen figured, the contralateral hemipenis originates at the level of subcaudal 44 (total paired subcaudals  =  55). Bogert (1940: 42) dissected another specimen with retracted hemipenes in which “The separate muscles attached to each fork are not straight, but are drawn back in a series of sinuous folds, as though there were insufficient space for both the hemipenis and the muscles within the sheath in the tail.” The contralateral hemipenis to the one in figure 28 is partially everted, but even so there was still evidence of the sinuous folding mentioned by Bogert.

Fig. 28.

The deeply divided, exceedingly long-lobed hemipenis of Pseudaspis cana (Linnaeus) (AMNH R-49948, right hemipenis). The contralateral (left) organ of this snake originates near the end of the tail at subcaudal 44 (there are 55 subcaudals total); although the left organ was partially everted, its retractor muscle retained a series of sinuous folds that seemed to straighten somewhat after being freed from membranous tissue. The branches of the sulcus spermaticus orbit each lobe several times; the sulcus orientation is “centrifugal-orbital” following the classification proposed herein. (The orbiting nature of the sulcus is not clearly evident in a partially everted organ illustrated by Cope, 1895: pl. 24, figs 9, 9a.)

Folding of the retractor muscle seems commonplace in still another African genus, Prosymna. Battersby (1951: 829) wrote:

It was noted, whilst sexing the two specimens [of Prosymna pitmani, sp. n.] that the retractor muscles were very large and lay in many convolutions from the hemipenis to practically the tip of the tail. In the paratype the right hemipenis is evaginated, the convolutions are straightened out, and the muscle is so much stretched that it is threadlike. Examination of several males of Prosymna ambigua stuhlmanni, P. melegris and P. somalica showed the same convoluted condition, but in specimens of P. sundevallii the retractor muscle was quite straight.

W.R. Branch (fide Broadley, 1980: 493–494) seems to have confirmed that a straight retractor muscle in Prosymna is correlated with the relatively short hemipenes of P. sundevallii subspp., P. angolensis, and P. bivittata; the straight retractor muscle in P. frontalis “is permitted by the relatively long tail of this species.”

Bogert (1940: 43) suggested that the sinuous folding of the retractor in Pseudaspis cana “is comparable to what must exist in Prosymna, the remarkable ‘telescoped’ hemipenis of which is longer when everted than the tail (vide Schmidt, 1923…).” Bogert was referring to Schmidt's description of a specimen of Prosymna ambigua bocagii (AMNH R-12145). However, dissection of that specimen by Myers shows a small Z-shaped fold in each retractor and a large Z-shaped fold situated posteriad in the left retractor. Although the last mentioned fold would have to contribute significantly to the unfolding and lengthening of the still retracted left hemipenis, there is no sinuous folding along the length of that retractor as expected by Bogert. Instead, the hemipenis itself is remarkably folded!

Figure 29A shows the everted hemipenis of Prosymna ambigua bocagii as illustrated by Schmidt; the tip of the organ was broken off and stored with the specimen in a vial. Figure 29B shows the same specimen as recently dissected, with an insert at the lower right showing a greatly enlarged view of the broken tip of the right hemipenis. The folding of the uneverted left hemipenis may be a bit more complex than “accordionlike,” but further dissection would be necessary to unfold or accurately describe it. The place of insertion of the retractor muscle on the uneverted hemipenis is not clear; a long area is enveloped in membranous and tough connective tissue and there seemed no compelling reason for destructive dissection of the only available specimen.

Fig. 29.

The African genus Prosymna includes examples of extreme hemipenial diversity. A. Prosymna ambigua bocagii Boulenger, a snake with a slender, single (unforked) hemipenis longer than the tail when everted. Drawing of AMNH R-12145, reproduced from Schmidt (1923: 89). B. Same specimen as above, with tail opened, showing complex “accordion” folding of the uneverted contralateral hemipenis. The retractor muscles originate threadlike at the very tip of the tail, both with a small Z-shaped fold toward the hemipenis. Notice at the bottom side that the everted right hemipenis has the tip broken off. Inset at top: Greatly enlarged view of the folded part of the inverted hemipenis. Inset at lower right: Enlarged view of the tip of the broken right hemipenis, showing ridgelike structures at the terminus. C. Prosymna sundevallii (A. Smith) hemipenes showing peculiar anastomosing flouncing, from Broadley (1980: pl. IIA). Abbreviations: AG, presumed small anal gland; HP, hemipenis; Mus, termini of left and right retractor muscles; MusZ, large Z-shaped fold in left retractor muscle; SS, sulcus spermaticus.

The divergence between the hemipenes of Eutrachelophis bassleri and E. steinbachi—along with the several extremely different hemipenes discussed and illustrated above—are indeed much more elaborate than would seem necessary for the “simple function of gamete transfer.” Such unexpected, unpredictable variation strikes a discordant note in the usual systematic utility of hemipenial data, in which genitalic divergence normally has inherent high stability, but such stability “may be overridden at times by direct selection on the organs themselves or pleiotropic events” (Arnold, 1986: 263). Truly extreme genitalic divergence appears remarkable when it occurs, but is not by itself sufficient reason to generically separate close relatives—particularly sister species.

Eberhard's Thesis

In 1985, William Eberhard called to attention an evolutionary trend so pervasive that many taxonomists were led to wonder, How has this been overlooked for so long? Eberhard postulated that male genitalia both of invertebrates and vertebrates diverge rapidly relative to other characters. Male genitalia do not ordinarily show species-specific differences in animals with external fertilization, but the situation is exactly the opposite for those species that are dependent on internal fertilization. Eberhard (1985) dwelt on the topic of rapid divergence in genitalia of the latter. In a broad overview of animal genitalia, Eberhard considered a variety of hypotheses related to copulation and diversification of genitalia (e.g., lock and key, male conflict, pleiotropism, etc.); he admitted that various factors were likely to be operative in different groups, but concluded that sexual selection was the most likely general explanation.

Eberhard has not been entrenched in the idea of selection by female choice and his later papers (e.g., 2004, 2009, and 2010) continue to explore topics as diverse as “antagonistic coevolution,” “cryptic female choice,” and “static allometry” of genitalia. It is a complicated and active field of research not to be summarized here, but the citations in Eberhard's papers provide a good introduction to the literature. Eberhard's thesis seems to have been generally accepted as a truism, with little or no opposition. The burgeoning literature mainly seeks explanations. It follows that the bulk of this literature stems from arachnologists and entomologists, since the intricate and diverse nature of genitalia in these huge groups have long been known.

Selection somehow involving sex (not merely female choice), or a combination of sexual and natural selection, often is invoked in the new literature, along with tests and examinations of the premises involved. Among the latter, Huber (2003: 69) concluded: “The taxonomic evidence in favor of rapid evolution and species-specificity seems overwhelming,” but he called for rigorous examination and implied that molecular phylogenies are best suited “to test relative rates of evolution in different character systems.”

Arachnologist Jonathan Coddington, in an early review of Eberhard's book, made the following observation:

Species specific differences between hemipenes of closely related snakes usually can be found without difficulty; paradoxically, possible convergence in hemipenial features of unrelated snakes are sometimes more evident (e.g., as mentioned herein, resemblances between the hemipenes of Rhadinaea vermiculaticeps and Rhadinella schistosa, and between those of Eutrachelophis steinbachi and Mehelya poensis).

The herpetological literature recently contributed to the complexity of the selection problem with a paper by King et al. (2009:110), who suggest that “sexual conflict over copulation duration may have shaped the evolution of hemipenis morphology, favoring more elaborate organs in species in which a long duration of copulation is especially beneficial to males, despite the associated cost to females.” Two species were tested and compared, Thamnophis sirtalis and T. radix. Copulation duration was shorter in T. sirtalis, which has a simple subcylindrical organ that is more easily disengaged by females. Duration was longer in T. radix, which has a larger, bilobed (T-shaped) hemipenis that is hard to disengage, causing the male to be dragged if the female moves. The difficult job of associating benefits to females on one hand and males on the other is discussed by the authors; their comments on evolutionary plasticity in hemipenis morphology of New World natricines is a reasoned judgment deserving consideration.

But, whatever the selection processes that are in play among different animals, the underlying mechanisms for change are less well understood. Eberhard's thesis of genitalic divergence includes evolutionary changes that (from a taxonomic perspective) appear relatively stable and informative on one hand and, on the other hand, seemingly inexplicable changes that are discordant and uninformative. Referring to the “extravagance or apparently superfluous complexity of many genitalia,” Eberhard (1985: 81–82) suggested “the potentially cumulative nature of complex modifications mean that there is no predictable upper limit to genitalic complexity” [emphasis added]. The last point is perhaps exemplified by the extreme hemipenial divergences described and illustrated herein for Eutrachelophis, Enulius, and Prosymna.

Are such novelties to be explained by natural selection of random mutations? Or can rapidly increasing knowledge of the regulation and expression of Hox genes provide a more direct explanation, at least in some cases?

Hox Gene Expression

Studies of evolutionary development were unified by last century's discoveries of Hox genes, their chromosomal locations, methods of their regulation and expression, and their universal taxonomic distribution among metazoans. These genes regulate embryonic development of all vertebrates and were crucial to the evolutionary origin of external genitalia (note: hemipenes are developmentally external)—an event linked to terrestrialism and internal fertilization. The earliest genital appendage is thought to have occurred either at the base of the tetrapods or at the base of the amniotes (Kondo et al., 1997; Perriton et al., 2002: 42–43). There are close parallels in development of the genital tubercle in rodents and reptiles. And the past few decades have witnessed an accumulation of data encompassing the Squamata (e.g., Cohn and Tickle, 1999; Di-Poi et al., 2009, 2010; Polly et al., 2001; Woltering et al., 2009); ongoing work involving hemipenes is awaited with interest.

Cohn and Tickle (1999) invoked complex collinear expression patterns of Hox genes to describe the developmental basis of limblessness and axial patterning in snakes—in part by homeotic mutations (transformations) along the body axis. Cohn (2002: 508) noted that “Although Hox genes are generally associated with regionalization of the primary body axis of the embryo, they are also involved in … secondary axes [including legs and genitalia].” He went on (loc. cit.: 509) to say that “Undoubtedly, morphological evolution does not occur solely by homeosis, and as the field develops, we can expect to discover more fine-scale anatomical changes that are the results of evolving Hox gene regulation.”

Although the evidence is not yet here, we suggest that the occasionally dramatic differences between hemipenes of closely affiliated snakes may one day be explained by fine-scale Hox gene regulation. At the moment, it seems to be the best guess we have to account for anatomical structures that seem otherwise inexplicable. Meanwhile, we look forward to the availability of molecular data for testing hypothesized relationships.