520 



Fishery Bulletin 103(3) 



Results 



Sequence variation and divergence 



An approximately 1.4-kb gene region was amplified 

 and sequenced from 93 samples representing 35 shark 

 species. Fifty-seven of the 93 individuals had unique 

 haplotypes (Table 1, Fig. 1). An alignment of these hap- 

 lotypes with several outgroups with the SAM algorithm 

 resulted in a 1510 position consensus alignment after the 

 introduction of gaps. Of these 1510 aligned positions, 717 

 positions were variable and 543 were parsimony informa- 

 tive. Transition outweighed transversion substitutions 

 by a factor of 4.27. Considering only phylogenetically 

 informative sites within the ingroup, we found that 

 nucleotide composition did not differ significantly among 

 haplotypes (A: 35.9%, C: 21.9%, G: 16.9%, T: 25.3%; 

 X 2 =175.6, P=0.39). 



Phylogenetic analysis 



Unweighted parsimony analysis produced 24 equally 

 parsimonious trees of length 2733 (CI=0.39, RI=0.74) 

 that differed primarily in the relationships among haplo- 

 types within species (not shown). Neighbor-joining anal- 

 yses produced nearly identical topologies regardless of 

 the distance metric used. When differences were noted, 

 they often involved trivial placements of individual vari- 

 ants within species or the placement of branches that 

 were poorly supported by bootstrap analyses regard- 

 less of the reconstruction method employed. For this 

 reason, we present phylogenetic hypotheses generated 

 by neighbor-joining, using p-distances as a surrogate 

 for all analyses. 



Most clades containing multiple haplotypes within 

 species were highly supported by bootstrap analyses. Of 

 16 species represented by more than a single sequence, 

 15 were recovered as monophyletic groups in 100% of 

 1000 bootstrap replicates (Fig. 1). Sequence divergence 

 within species was generally trivial compared to among- 

 species divergences. For example, sequence divergence 

 among haplotypes within species of Carcharhinus dif- 

 fered by approximately two orders of magnitude from 

 that among species within the genus (average p-dis- 

 tance of 0.05% and 4.16%, respectively). The exception 

 involved haplotypes observed within C. plumbeus that 

 were supported as monophyletic by fewer than 70% 

 of 1000 bootstrap replicates in MP and NJ analyses. 

 Interestingly, a sister group relationship between C. 

 plumbeus and C. altimus was highly supported by boot- 

 strapping, and average sequence divergence within spe- 

 cies (0.14%) was only about one-third of that observed 

 between these two (0.43%). 



Some higher order relationships were recovered with 

 high bootstrap support. Notably the Carcharhiniformes 

 were strongly supported as monophyletic, as were the 

 families Sphyrnidae and Triakidae. The family Car- 

 charhinidae was poorly supported as monophyletic, al- 

 though a group that included Negaprion, Prionace, and 

 all Carcharhinus was observed in a large number of 



bootstrap replicates. Carcharhinus was paraphyletic in 

 the NJ topology, and Negaprion was nested within the 

 genus, but this relationship received little support from 

 bootstrapping. The Lamniformes were monophyletic 

 and strongly supported by bootstrapping. Within this 

 order, only the family Lamnidae received strong sup- 

 port, whereas support for a monophyletic Alopidae was 

 moderate. The order Hexanchiformes was recovered as 

 a monophyletic group; however bootstrap support for 

 this grouping was low. 



Discussion 



Our goal was to assess whether the 12s-16s region of 

 the shark mitochondrial genome contained sufficient 

 genetic variation and phylogenetic signal to be useful 

 in species identification. Of the 35 species examined, 

 6 species were each represented by a single individual, 

 and 16 of the remaining 29 species contained variants 

 at the mtDNA locus examined. Importantly, all within- 

 species variants formed strongly supported monophyletic 

 groups concordant with morphologically based species 

 descriptions. Intraspecific variability was low in rela- 

 tion to interspecific divergence at this locus and in no 

 instance was a paraphyletic relationship between spe- 

 cies observed. The combination of limited intraspecific 

 variability combined with sufficient between-species 

 divergence indicates that this locus is suitable for spe- 

 cies identification. 



Two exceptions to this generalization of low within 

 versus large between-species differentiation exist in our 

 phylogenetic hypothesis — one involving the sister spe- 

 cies pair C. plumbeus and C. altimus. In an alignment 

 of mitochondrial sequences from these species, only 5 

 or 7 transition substitutions were observed across ap- 

 proximately 1.4 kb of sequence data. Interestingly, Heist 

 and Gold (1999) included these two taxa in their cyto- 

 chrome^ RFLP analysis, and again, Atlantic samples 

 of C. plumbeus and C. altimus differed by only a single 

 transition in 395 basepairs (0.25%), and there were 

 more substitutions observed between Atlantic and Pa- 

 cific C. plumbeus than between Atlantic samples of C. 

 plumbeus and C. altimus (Table 2 in Heist and Gold 

 1999). The next most closely related pair of taxa in 

 our phylogenetic hypothesis comprised two other Car- 

 charhiniforms. C. longimanus and C. obscurus, a taxon 

 pair differing by approximately 1.44% sequence diver- 

 gence, compared with an average of 0.06% within taxon 

 diversity. These two taxa were considered by Shivji et 

 al. (2001) while developing a multiplex PCR assay for 

 six commercially important pelagic species. Specifically, 

 assays developed to diagnose C. obscurus could not 

 discriminate between C. obscurus and C. longimanus, 

 two closely related species in our phylogenies. The C. 

 plumbeus and C. altimus species pair was not consid- 

 ered by Shivji et al. (2001); thus no comparison to the 

 Heist and Gold (1999) cytochrome-6 sequence/RFLP or 

 the 12s-16s data set presented in our study was pos- 

 sible. We are currently analyzing additional samples, 



