Phylogenetic relationshiPs of the genus SibynophiS ( serPentes : colubroidea )

We present the results of the first molecular analysis of the phylogenetic affinities of the Asian colubroid genus Sibynophis. We recovered a sister-group relationship between Sibynophis and the New World Scaphiodontophis. Although Liophidium sometimes is associated with these genera, the relationship is distant. Morphological characters that Liophidium shares with Sibynophis and Scaphiodontophis are resolved as homoplasies that probably reflect the similarities of their specialized feeding habits. The traditional subfamily Sibynophiinae is elevated to the family-level, and Scaphiodontophiinae is placed in its synonymy.


IntroductIon
The genera Liophidium, Sibynophis, and Scaphiodontophis occur on three distinct landmasses-Madagascar, Asia, and Central America, respectively.Despite their isolation, these snakes long have been thought to be closely related to each other.In each genus, the dentary bears a peculiar posterior dentigerous process that is completely detached from the compound bone and teeth are numerous and closely set (Duméril et al., 1854;Boulenger, 1890Boulenger, , 1896)).Duméril et al. (1854) were the first authors to place the four species that share these morphological characteristics in the subgenus Enicognathus of their genus Ablabes.Later, Boulenger (1890) substituted Enicognathus, preoccupied, with Polyodontophis that he recognized as a genus.He added four additional species from India and southeastern Asia to Polyodontophis, as well as P. comorensis (= Liophidium mayottensis) and P. sumichrasti from the Comoros Islands and Central America, respectively.Boulenger (1896) further expanded Polyodontophis with three additional species, and described the genus Liophidium, to accommodate L. trilineatum from Madagascar.Polyodontophis was posteriorly synonymized with Fitzinger's (1843) Sibynophis when Stejneger (1910) noted the latter genus has priority over the former, an arrangement followed by all subsequent authors (e.g., Dunn, 1928;Smith, 1943;Taylor & Smith, 1943).Dunn (1928) partially formalized Duméril's and Boulenger's perspectives by erecting the subfamily Sibynophiinae to accommodate the Asian and Central American species of the genus Sibynophis, but he did not include the Malagasy taxa.Although Dunn did not explicitly refer to the Malagasy species of Sibynophis, he implicitly considered them to be unrelated to the Asian and Central American species.According to Dunn (1928), both Asian and Central American Sibynophis retain a single-lobed hemipenis with a single sulcus spermaticus-a condition strikingly distinct from the bilobed hemipenis with a forked sulcus that occurs in the Malagasy species of Sibynophis.Taken together, both dentitional and hemipenial similarities between the Asiatic and New World species were so compelling that Dunn decided to consider them congeneric (Dunn, 1928: 20).It was only after Taylor & Smith (1943) that the New World species of Sibynophis were allocated in a distinct genus Scaphiodontophis.
After Dunn (1928), Bogert (1940) briefly discussed the hemipenial differences between the Malagasy Sibynophis torquatus and its Asiatic and Central American congeners, but he did not remove the taxon from Sibynophiinae.Following Bogert (1940), Leviton & Munsterman (1956) erected the genus Parasibynophis to accommodate the Malagasy and Comoran species Sib.torquatus, Sib.rhodogaster, and Sib.mayottensis, distinguishing them from the Asian and New World taxa on the basis of 11 osteological and hemipenial characters.Like Bogert (1940), these authors considered the mandibular characters to be of greater phylogenetic importance than the hemipenial features, reinforcing the allocation of all three genera Scaphiodontophis, Sibynophis, and Parasibynophis into the subfamily Sibynophiinae.Later, Guibé (1958) synonymized Parasibynophis with Liophidium, an arrangement followed by subsequent authors (e.g., Underwood, 1967;Williams & Wallach, 1989;Cadle, 1999).
Several more recent authors recognized either the subfamily Sibynophiinae or tribe Sibynophiini (e.g., Kuhn, 1967;Underwood, 1967;Smith et al., 1977;Ferrarezzi, 1994;Zaher, 1999).Further, Taylor & Smith (1943), Leviton & Munsterman (1956), Underwood (1967), Ferrarezzi (1994), and Zaher (1999) explicitly recognized either the Sibynophiinae or Sibynophiini as a natural assemblage of hingedteeth snakes that includes Scaphiodontophis and Sibynophis but excludes Liophidium.Underwood (1967: 91) stated the issue with clarity: "this subfamily [Sibynophinae] as proposed by Dunn included the Madagascar species now put into Liophidium; the relationships of these latter snakes lie elsewhere but the remaining two genera Sibynophis and Scaphiodontophis are certainly related and systematically rather isolated." Ferrarezzi (1994: 90) provided a diagnosis for the tribe Sibynophiini that, according to the author, includes two putative synapomorphies: "posterior process of the dentary free and elongate; maxillary teeth small, robust and flattened laterally."Although shared by Scaphiodontophis and Sibynophis, these two characters also are present in Liophidium, in which they have evolved homoplastically.Zaher (1999: 28) further documented that Sibynophis and Scaphiodontophis share an additional putative hemipenial synapomorphy-viz., a sharply curved, U-shaped convolution of the sulcus spermaticus in the proximal region of the hemipenial body.
More recently, Pyron et al. (2011) erected a new subfamily of colubroid snakes, the Scaphiodontophiinae, to accommodate the New World genus Scaphiodontophis.Their analysis also successfully tested the phylogenetic position of Liophidium, which appears as a member of their subfamily Pseudoxyrhophiinae, and not as the sister-group of Scaphiodontophis.However, monophyly of the traditional subfamily Sibynophiinae, including the genera Sibynophis and Scaphiodontophis (sensu Zaher, 1999), requires testing because Sibynophis has not been sampled in any of the recent molecular phylogenetic analyses.Here, we test this hypothesis by adding Sibynophis collaris to Pyron et al.'s (2011) data matrix, and discuss the derived osteological and hemipenial characters shared by these genera.

DNA from
Sibynophis collaris was extracted from liver following the protocol of Hillis et al. (1996).Sequences for two nuclear protein-coding genes were amplified via polymerase chain reaction (PCR): oocyte maturation factor (c-mos) and recombinationactivating gene 1 (rag1).To amplify c-mos we used the primers described in Zaher et al. (2009); and for rag1 we used the primers Mart.FL1 (5' AGC TGC AGY CAR TAY CAY AAR ATG TA 3') and Amp.R1 (5' AAC TCA GCT GCA TTK CCA ATR TCA 3') as described by Chiari et al. (2004).PCRs were performed using standard protocols with the addition of 0.4% of Triton X-100 (Sigma-Aldrich, Saint Louis, MO, USA) and annealing temperature of 54°C.
PCRs were purified with shrimp alkaline phosphatase and exonuclease I (GE Healthcare, Piscataway, NJ).Sequences were processed using a BigDye Terminator cycle sequencing kit in an ABI 3700 sequencer (Applied Biosystems, Foster City, CA, USA).Both strands were sequenced and checked.The consensus of the two strands was constructed using Geneious version 5.4.6 (Drummond et al., 2011).
The voucher of Sibynophis collaris used in this study was deposited at the Royal Ontario Museum (ROM 25618).Sequences of rga1 and c-mos were submitted to GenBank, receiving accession numbers JQ994296 and JQ994297, respectively.One sequence of Sibynophis chinensis for cytochrome b (cytb) was downloaded from GenBank (accession JN571580).We assumed monophyly of the genus Sibynophis and this allowed the creation of a combined terminal (c-mos and rag of S. collaris and cytb of S. chinensis).
We included the sequences of Sibynophis in the matrix of Pyron et al. (2011), while preserving the existing alignment, by using the command "--add" available in MAFFT version 6.864 (Katoh et al., 2002).We conducted a maximum likelihood analysis using RAxML 7.2.8 (Stamatakis, 2006).As performed by Pyron et al. (2011), the concatenated data were partitioned by codon position and the GTRGAMMA model was used for all partitions.The rapid bootstrap algorithm was employed, and 500 non-parametric bootstrap replicates were performed using the cluster hosted at the Laboratório de Alto Desempenho -Pontifícia Universidade Católica do Rio Grande do Sul (LAD-PUCRS).This approach allowed 100 complete searches because it used the tree generated at every fifth bootstrap replicate as the starting tree for an independent LSR (Lazy Subtree Rearrangement) swapping.Nodal support was assessed through bootstrapping (BS).
Cleared-and-stained skulls were prepared following the technique of Dingerkus & Uhler (1977) in which cartilage is stained with Alcian blue, bone with Alizarin red, and tissue is macerated with Trypsin.Skeletal nomenclature follows that of Cundall & Irish (2008).Methods for hemipenial preparation and terminology followed those of Zaher (1999) and Zaher & Prudente (2003).
FIgure 1: Summary tree showing the major clades of Caenophidia as recovered in the maximum likelihood analysis of Pyron et al.'s (2011)

results
Our tree-search procedure in RAxML produced a topology with a score of lnL = -376071.295841.A summary of the tree topology is given in Figure 1, and the full tree is provided as supplementary material (Appendix S1).Although we retrieved the same well-supported clades (BS > 70%) as did Pyron et al. (2011), the relationships at poorly supported nodes frequently differed (e.g., affinities between elapoid families sensu Zaher et al., 2009).

dIscussIon
According to the phylogenetic affinities recovered in our molecular analysis (Fig. 1), Scaphiodontophis and Sibynophis share four putative morphological synapomorphies that are absent in Liophidium and all remaining pseudoxyrhophiines.These are, as follow: (1) distally broadened, plate-like maxillary process of the palatine with an anterior edge projecting from the level of the anterior tip of the palatine and covering completely the palatine process of the maxilla (Fig. 2); (2) maxilla projects posteriorly to the maxillaryectopterygoid contact, forming a significantly long "dentigerous process" posterior to the ectopterygoid process of the maxilla (Fig. 2); (3) choanal process of the palatine absent (Fig. 2); and (4) sharply curved U-shaped convolution of the sulcus spermaticus in the proximal region of the hemipenial body (Fig. 3).
Alternately, five morphological characters found in Scaphiodontophis and Sibynophis appear to have evolved homoplastically in Liophidium.These characters include: (1) numerous, closely set teeth (Figs. 2, 4); (2) compound bone anterior to the mandibular fossa with a long, tubular and dorsally curved shape (Fig. 4); (3) splenial and angular bones reduced (Fig. 4); (4) mandibular fossa reduced and restricted to the posterior one-fifth of the compound bone (Fig. 4); and (5) posterior dentigerous process of the dentary elongated and detached from the compound bone (Fig. 4).These homoplastic characters likely reflect the specialized durophagous diet of these snakes (Savitzky, 1981(Savitzky, , 1983)).Cadle (1999), following Savitzky's (1981) lead, found hinged and peculiarly shaped teeth in Pseudoxyrhopus, Exallodontophis, Heteroliodon, and Pararhadinaea; all of these Malagasy genera are closely related to Liophidium and known to feed on durophagous prey.Cadle (1999) also pointed out that hinged teeth occur in some African and Asian snakes that feed on hard prey (e.g., Lycophidion, Mehelya, Chamaelycus, Tropidonophis).The apparently widespread geographical distribution of hinged teeth represents additional evidence that this peculiar tooth morphology has arisen independently several times.an expanded Lamprophiidae, and a highly speciose Colubridae.We acknowledge the fact that this scheme promotes taxonomic stability for the long-standing name Colubroidea.However, such traditional taxonomic hierarchy no longer accommodates new knowledge of phylogenetic affinities within the group appropriately (Vidal et al., 2007;Kelly et al., 2009;Zaher et al., 2009).
As knowledge of the phylogenetic affinities of colubroideans improves, additional taxonomic changes that create a certain amount of instability will be required to accommodate the newly resolved relationships (Lawson et al., 2005;Vidal et al., 2007;Zaher et al., 2009).Such taxonomic adjustments, provided that they are stated explicitly (Zaher et al., 2009), are necessary to incorporate the diversity and newly discovered monophyletic groupings of the colubroidean radiation.
As pointed out by Frost et al. (2006: 143) "We expect that regulated nomenclature will increasingly be pushed toward the terminal taxa and that unregulated taxa will increasingly be rankless.The reason for this is that there really is a practical limit to the number of ranks that workers are willing to use … So, our observation is that sociological pressures will push workers towards ever smaller families … Regardless, we think that this process will correspond with enormous progress in phylogenetic understanding."Zaher et al.'s (2009) new taxonomic scheme represents a step in that direction for the largest, most speciose group of snakes.

AcKnowledgMents
The authors wish to thank Charles Myers, Darrel Frost and David Kizirian (AMNH), Ronald Nussbaum and Greg Schneider (UMMZ), William Duellman and Linda Trueb (KU) for their support and for allowing us to prepare and analyze specimens from the collections under their care.We are grateful to the staff of the Laboratório de Alto Desempenho, Pontifícia Universidade Católica do Rio Grande do Sul (LAD-PUCRS), who generously allowed access to the cluster under their care.We are also indebted to John Cadle, Linda Trueb, and Gustavo Scrocchi for their insightful review of a previous version of the work.Funding for this study was provided by the Fundação de Amparo à Pesquisa do
data set with Sibynophis sequences added (full topology in Appendix S1).Taxon names in bold indicate high-level taxa (higher than family).Numbers near nodes indicate bootstrap values for unnamed clades.Numbers in parenthesis after a taxon name indicate bootstrap values for that clade.