92 



Fishery Bulletin 93(1). 1995 



and T. albidus as separate species as well as T. pfluegeri, 

 T. angustirostris, and T. belone. 



Our data conflict with other aspects of current 

 istiophorid taxonomy at the generic level. Analysis 



6S IV 



5S 



5S IV 



20S 15V 



Istiophorus plaryplerus 



hrr Makaira 

 nigricans 



Tetrapturus albidus 



7, Tetrapturus audax 



IS 



65 



15S IV 



Tetrapturus 

 angustirostris 



Tetrapturus pfluegeri 



Z Tetrapturus belone 



4Jj- Makaira indica 



Xiphias gladius 



Coryphaena equiselis 



Figure 8 



Phylogeny of the Istiophoridae based on 78 informative nucle- 

 otide sites. The phylogram depicted is the single most parsimo- 

 nious tree identified by the exhaustive search option of PAUP 

 3.1 (Swofford, 1991). Xiphias gladius and Coryphaena equiselis 

 were specified as the outgroup. Length, consistency index, and 

 retention index are the following: L=152, CI=0.776, RI=0.721. 

 Circled numbers at nodes indicate the percentage of trials in 

 which a given partition between taxa is supported in 2,000 repli- 

 cations of the bootstrap analysis (Felsenstein, 1985). Character 

 state transformations were inferred by using the accelerated 

 transformation (ACCTRAN) option of PAUP: S=nucleotide tran- 

 sitions (A»G or C<=>T); V=nucleotide transversions (C/T<=>A/G). 

 Within the Istiophoridae transitions outnumber transversions 54 

 to 6. Neighbor-joining (Saitou and Nei, 1987) and UPGMA den- 

 drograms produced with Phylip 3.5 (Felsenstein, 1993) have the 

 same topology. Distance trees were constructed by using Kimura's 

 (1980) two parameter genetic distance, and by assuming a tran- 

 sition to transversion bias of 9:1. 



of cytochrome b does not support the monophyly of 

 the genus Makaira. The black marlin, Makaira 

 indica, appears to be the sister group of a clade con- 

 taining all other istiophorids, while the blue marlin, 

 M. nigricans, is sister group of the sailfish, /. 

 platypterus. The most parsimonious tree that 

 contains a monophyletic Makaira is six steps 

 longer than the shortest tree overall (158 ver- 

 sus 152), and on the most parsimonious tree, 

 the M. nigricans-I. platypterus node is strongly 

 supported by bootstrap analysis (85%). 



Cytochrome b provides good resolution of the 

 relationships of the genera of the tribe Thunnini 

 (Auxis, Euthynnus, Katsuwonus, and Thunnus). 

 According to the nucleotide data the nine 

 Thunnini species sequenced in this study com- 

 prise two clades, one consisting of the genus 

 Thunnus and one containing the other genera: 

 Auxis, Euthynnus, and Katsuwonus. This dis- 

 tinct split in the Thunnini was proposed by 

 Kishinouye in 1923 and is consistent with the mor- 

 phological hypothesis of Collette et al. (1984). 

 Support for the monophyly of the Thunnus clade 

 is particularly robust; however, the relation- 

 ships within the genus cannot be resolved with- 

 out the inclusion of both Thunnus tonggol and 

 Thunnus atlanticus which were not sequenced 

 in this study. The number of substitutions sepa- 

 rating T. thynnus from T. maccoyii (<0.5% se- 

 quence divergence) are small considering their 

 status as separate species. 



Discussion 



Interfamilial relationships and the limits 

 of the Scombroidei 



Throughout this analysis, we have focused on 

 the long-standing controversy over the limits 

 of the Scombroidei and, particularly, whether 

 billfishes are scombroids. Cytochrome b appears 

 to be informative on this issue. In the two phy- 

 logenetic analyses that emphasize the more 

 slowly evolving characters (see Figs. 4 and 7), 

 the most parsimonious tree topology is clearly 

 most consistent with the hypothesis that bill- 

 fish are not scombroids: in each case, one or 

 more nonscombroids share a common ancestry 

 with the scombrid-gempylid-trichiurid clade to 

 the exclusion of billfishes (Table 4). Therefore, 

 according to the criterion of parsimony, the 

 nonscombroid hypothesis is superior to the 

 scombrid subgroup and to the scombrid sister 

 group hypotheses. However, in our opinion, the 



