102 



W. WUSTER£T/U. 



PLATE 1 



Bothrops caribbeus © R.S. Thorpe 



1999), using 580 b.p. of cytochrome b sequence, found B. caribbaeus 

 and B. lanceolatus to the sister species of the South American 

 populations of the B. atrox complex. However, the study included 

 only a limited sampling of South American members of the B. atrox 

 complex, and did not include representatives of B. asper from 

 Central America. 



The aim of this paper is to explore in more depth the origin of the 

 Antillean Bothrops, and its implications for other fields, using an 

 expanded dataset of more sequence information from a larger number 

 of potentially related species. 



MATERIALS AND METHODS 



We obtained tissue (ventral scale clippings or tail tips) and/or blood 

 samples from species representing the principal clades within the 

 genus Bothrops (including Bothriopsis), as well as the closely 

 related Bothrocophias - see Wiister et al. (in press). We also 

 included samples of the B. asper-atrox complex from around the 

 coast of South and Central America, as these have been considered 

 to be potential founder populations from which the ancestor of 

 Antillean Bothrops could have arisen. For outgroup rooting, we used 

 sequences of Bothrops alternatus and Bothrocophias microphthal- 

 mias. Two regions of the mitochondrial DNA molecule were amplified 

 using the polymerase chain reaction (PCR): a 767 base pair (bp) 

 section of the gene for cytochrome b (cyib), and a 900 bp region of 

 the gene for NADH dehydrogenase subunit 4 (ND4). Details of 

 primers and laboratory protocols are given in Pook et al. (2000). 



Sequences were aligned by eye against published Bothrops 

 sequences (Puorto et al., 2001). In order to test for the presence of 

 saturation of certain categories of substitution, we calculated Tamura- 

 Nei distances between all samples. This takes into account deviations 

 from equal base compositions and differences in substitution rates 

 among nucleotides. We then plotted unadjusted p-distances for 

 transitions and transversions, and for the three codon positions 

 separately, against Tamura-Nei distances. A decline in the rate of 

 accumulation of individual categories of substitution with increased 

 Tamura-Nei distances indicates saturation of that substitution category. 



We checked all sequences for insertions, deletions or the presence 

 of stop codons. Any of these would have indicated that the sequences 

 represent nuclear insertions of the mitochondrial genes (Zhang and 

 Hewitt, 1996). The sequence data were assayed for the presence of 

 a significant phylogenetic signal by means of the gl tree skewness 



Bothrops lanceolatus © D. Warrell 



statistic (Hillis and Huelsenbeck, 1992), calculated from 100,000 

 trees randomly generated by PAUP* 4.0b8 (Swofford, 2001). 

 Sequence divergences between clades were estimated using the 

 program Phyltest (Kumar, 1996). 



We analysed our sequence data using both maximum parsimony 

 (MP) and maximum likelihood (ML) as optimality criteria. Using 

 multiple optimality criteria should identify those parts of a 

 phylogenetic tree that are supported independently of the optimality 

 criterion used. Such nodes should inspire greater confidence than 

 nodes that are unstable and vary depending on method of analysis. 

 All analyses were carried out using the program PAUP* 4.0b8 

 (Swofford, 2001). 



For MP analyses, we selected Bothrops alternatus and B. micro- 

 phthalmia as outgroups. We employed the heuristic search algorithm 

 of PAUP* 4.0b8, using TBR branch swapping and 100 random 

 addition sequence replicates. The analysis was carried out on the 

 unweighted data only. 



The extent to which individual nodes on the tree were supported 

 by the data was assessed using bootstrapping and Bremer (1994) 

 branch support. Non-parametric bootstrap was implemented using 

 heuristic searching, 1000 replicates, TBR branch swapping and 10 

 random-addition-sequence replicates per bootstrap replicate. Bremer 

 branch support for individual nodes was calculated through the use 

 of the converse constraint option of PAUP*. 



For ML analyses, we identified the most appropriate model of 

 sequence evolution through the use of the MODELTEST software 

 (Posada & Crandall, 1998). A first ML search was run, using 

 heuristic searching, a neighbour-joining starting tree and TBR branch 

 swapping, and the sequence evolution parameters identified by the 

 Modeltest software. These parameters were then re-estimated from 

 the resulting ML tree, and a further search run using these re- 

 estimated parameters. This was repeated until further estimates 

 yielded no further changes of parameter values or tree likelihood 

 scores. Bootstrap analysis involved 100 replicates, using NJ starting 

 trees and NNI branch swapping. 



An important consideration of any proposed scientific hypothesis 

 is whether the data supporting it can reject alternative hypotheses 

 with statistical significance. In other words, do the data allow us to 

 reject the null hypothesis that differences in tree optimality between 

 the optimal tree and trees consistent with alternative hypotheses are 

 due to random chance? In the case of the Antillean Bothrops, we 

 tested the following alternative phylogenetic hypotheses: (i) non- 

 monophyly of the Antillean Bothrops, i.e., the Antillean populations 

 of Bothrops result from separate colonisation events; (ii) non- 



