PHYLOGENY OF SOME MIDDLE AMERICAN PITVIPERS BASED ON A CLADISTIC ANALYSIS OF MITOCHONDRIAL 12S AND 16S DNA SEQUENCE INFORMATION

The cladistic relationships of several Middle American pitvipers representing the genera Bothrops (sensu stricto), Bothriechis, Cerrophidion, Lachesis and Porthidium were determined using mitochondrial 12S and 16S DNA sequence information. Maximum parsimony analyses were performed using PAUP on aligned sequences that included published information for related taxa. Two sets of analyses were conducted: one disregarding gaps in the aligned matrix, and another with gaps treated as a fifth base. When gaps were excluded resolution declined, although the general arrangement of the taxa changed little. A consistent relationship was the grouping of ((Porthidium, Bothriechis) Lachesis). The placement of Lachesis, as nested within other bothropoid genera, is only partially supported by results of other authors. The arrangement of Crotalus, Bothrops and Cerrophidion was ambiguous when gaps were discounted. In both trees, Agkistrodon was basal to the New World forms. The remaining genera, Trimeresurus (Protobothrops), Vipera, Azemiops, and Coluber, were uniformly distant to the former taxa. Also of interest is the lack of close relationship, based on the DNA data here and elsewhere, between Bothrops and Porthidium. This is in striking contrast to results based on morphologic and allozymic analyses of previous studies. It is concluded that additional DNA sequence information from a larger sample of taxa will be necessary to better assess the phylogenetic relationships among Middle American and related pitvipers. The bothropoid pitvipers comprise a diverse and widespread assemblage of venomous snakes distributed from southern Mexico to southern Argentina. In the last decade, progress from systematic studies of pitvipers has led to descriptions of new species and the recognition of several new generic arrangements for those species formerly assigned to Bothrops (sensu lato). Detailed species accounts can be found in Campbell and Lamar (1989), whereas the most recent generic arrangement of New World pitvipers can be found in Campbell and Brodie (1992), and summarized in Campbell and Lamar (1992). The phylogenetic relationships among Neotropical pitvipers remain problematic. Although several studies using phenotypic character information (Crother et al., 1992; Werman, 1992; Gutberlet, 1998) and molecular data (Knight et al., 1992; Kraus et al., 1996; Cullings et al., 1997; Salomão et al., 1997; Vidal et al., 1997; Wüster et al., 1997) have been completed, many inconsistencies of phylogenetic inference persist (see Werman, 1998). Herein, we present a cladistic analysis of novel mitochondrial 12S and 16S DNA sequence information for some Middle American pitviper species, in conjunction with published sequences for related genera. Of primary concern is the relationship of Lachesis to the other New World pitviper genera and the relationship of Bothrops (sensu stricto) to Porthidium. Among DNA studies that include Lachesis (Kraus et al., 1996; Cullings et al., 1997; Vidal et al., 1997) there is no clear agreement as to the position of this genus relative to other pitviper genera. Lachesis is either a somewhat basal lineage, or is found nested among other bothropoid genera (see Werman, 1998). Regarding Bothrops and Porthidium none of the present DNA analyses place these together as sister lineages. This is curious because Werman (1992), based on a cladistic analysis of phenotypic information, regarded them as a terminal clade of recent divergence among the New World pitvipers. MATERIALS AND METHODS Tissue samples of five pitviper species, representing several Middle American genera, were used as DNA sources. Total genomic DNA samples (that included mitochondrial DNA) were isolated and purified from liver and/or skeletal muscle tissue using standard digestion (SDS-Proteinase K) and extraction (phenol-chloroform) techniques. Microgram quantities of genomic DNAs were obtained for the following taxa: Bothrops asper, Bothriechis rowleyi, Cerrophidion godmani, Lachesis muta, and Porthidium nasutum. Locality data and voucher information are available on request from the authors. Amplifications were carried out in a Thermolyne Amplitron thermal cycler utilizing established protocols with modifications specific for pitvipers (Knight and Mindell, 1993). Symmetrical amplification of the 12S sequences was accomplished with the L strand primer, 5'-AAACTGGGATTAGATACCCCACTAT-3', and the H strand primer, 5'GTACACTTACCTTGTTACGACTT-3'. The 16S sequences were amplified with the L strand primer, 5'-CGCCTGTTTATCAAAAACAT-3', and the H strand primer, 5'CCGGTCTGAACTCAGATCACGT-3' (Knight and Mindell, 1993). The cycle parameters for the 12S and 16S sequences were: 85o C, 5 min, followed by 30 cycles of 94o C, 35 sec; 50o C, 35 sec; 72o C, 1 min; with an extension on the last cycle of 72o C, 5 min, then 4o C, dwell/soak. The amplifications resulted in approximately 960 bp for the combined 12S and 16S gene fragments. Amplification products were either purified or ligated directly into Invitrogen pCR II TA cloning vectors following Invitrogen specific protocols. Ligation products were transformed into Invitrogen "One Shot" competent cells and plated on LB/Amp/X-gal media for colony selection. Plasmid DNAs were isolated from positive colonies using 5 Prime-3Prime "Perfect prep" miniprep kits. Purified plasmid DNAs were digested with EcoRI, to release the inserts, and size fractionated on 2% agarose gels. Plasmids positive for 12S and 16S inserts were collected and stored at -70o C. Sequencing effort included standard dideoxynucleotide termination methods (Sanger et al., 1977; Hillis et al., 1996) using CircumVent thermal cycle DNA sequencing kits (New England Biolabs) in combination with chemiluminescent Phototope detection kits (NEB). In addition, sequences were also determined with an automated Applied Biosystems Inc. (ABI) Prism DNA sequencer. In both cases, standard M13 forward and reverse sequencing primers were used. Sequences were determined by comparing both forward and reverse sequencing reactions. Approximately 410 nucleotide positions of the 12S gene and 550 nucleotide positions for the 16S gene were scored (e.g., 410 and 548 for P. nasutum 12S and 16S, respectively). Partial 12S and 16S DNA sequences for Coluber constrictor, Vipera ammodytes, Azemiops feae and Agkistrodon bilineatus were obtained from Knight and Mindell (1993). 12S sequence information for Trimeresurus (Protobothrops) mucrosquamatus was obtained from Genbank, accession # D31613 (Eguchi, unpublished, 1994). 12S and 16S sequences for Crotalus aquilus were also obtained from Genbank under the accession numbers L14373 and L14374, respectively (Knight et al., 1993). These six taxa were used for outgroup comparisons and rooting purposes in the cladistic analyses. Sequence alignments were accomplished by eye using a color coding scheme specific for each nucleotide. Insertions and deletions in the sequence matrix generated 975 total characters. Aligned sequences were subjected to phylogenetic analysis using PAUP, version 3.1.1 (Swofford, 1993). Maximum parsimony (MP) analyses were performed using random addition of sequences, tree-bisection-reconnection branch swapping, the MULPARS option with ACCTRAN optimization. Gaps were either treated as a fifth base or as missing information in the data matrix. Characters were not weighted. Unknown nucleotide information was designated as "n" and treated as missing information. Strict consensus trees were retained for multiple MP solutions. Bremer support values (Bremer, 1994), that identify how many extra steps on a particular branch are necessary to collapse the branch in a consensus tree of proximate parsimonious solutions, were used to assess nodal stability.Two analyses were performed: the first with gaps included as character information and a second where gaps were discounted. The rationale for these analyses was to determine the effect of gaps on cladogram resolution.