Finnerty and Block: Evolution of cytochrome b in the Scombroidei 



93 



most parsimonious trees alone do not constitute suf- 

 ficient evidence to reject these unfavored hypotheses. 

 The question we must ask is the following: How 

 unparsimonious are these hypotheses? 



In comparing the tree topologies that support each 

 competing hypotheses (Table 4), it is clear that our 

 data refute the notion that billfishes share a com- 

 mon ancestor with the Scombridae to the exclusion 

 of other scombroids (Gregory and Conrad, 1937; Berg, 

 1940; Fraser-Brunner, 1950; Collette et al., 1984; 

 Johnson, 1986). For example, the shortest trees sup- 

 porting a billnsh-scombrid clade are 13% longer than 

 the minimum-length tree based on inferred amino 

 acid sequence (Table 4). According to the cladistic 

 permutation test for nonmonophyly (Faith, 1991), 

 this length difference constitutes significant evidence 

 against the monophyly of scombrids plus billfishes. 

 The condition that billfishes and scombrids comprise 

 a monophyletic group is a requirement of both the 

 scombrid subgroup and scombrid sister group hypoth- 

 eses. Therefore, according to the cytochrome b data, 

 we reject these two hypotheses. 



The cytochrome b data clearly support the third 

 hypothesis, that billfishes are not scombroids, though 



not as strongly as they refute the first two hypoth- 

 eses. According to the inferred amino acid sequences, 

 the shortest tree that supports scombroid monophyly 

 places billfishes as sister group to all other scom- 

 broids and is nearly 3% longer than the most parsi- 

 monious tree overall (145 versus 141 steps). This 

 length difference alone does not constitute signifi- 

 cant evidence against the monophyly of the Scom- 

 broidei according to a cladistic permutation test for 

 nonmonophyly (see Methods section; Faith, 1991). 

 However, as previously mentioned, there are three 

 amino acid characters that unite scombrids, 

 gempylids, and trichiurids with Sphyraena to the 

 exclusion of billfishes (amino acids 15, 16, and 169). 

 There are no characters that unite scombrids, 

 gempylids, and trichiurids with billfishes to the ex- 

 clusion of the putative outgroups. Our study is con- 

 sistent with the hypothesis that billfishes are most 

 closely related to some percoid lineage (Nakamura, 

 1983; Potthoff et al., 1986). The question of which 

 taxon is most closely related to billfishes remains 

 unanswered. On the basis of this evidence, we sup- 

 port a conservative definition of the Scombroidei, 

 including only the families Scombridae, Gempylidae, 



