FISHERY BULLETIN: VOL. 81, NO. 2 



SCOMBRINI 



SCOMBEROMORINI 



SARDINI 



THUNNINI 



GASTEROCHISMATINAE 



SCOMBRINAE 



FIGURE 16. — Diagram of relationships of scombrid fishes modified 

 from Collette and Russo (1979:fig. 1). 



S. Farris (following Farris 1970 and Farris et al. 

 1970). Infestation by a given copepod species was, 

 somewhat arbitrarily, considered primitive; absence, 

 specialized. A transformation series was used to indi- 

 cate decreasing amounts of parasitism by a given 

 copepod species across a matrix of scombrid genera. 

 The Wagner tree was rooted at Rastrelliger, one of the 

 most primitive members of the Scombrinae. The 

 resulting Wagner tree (Fig. 17) shows major differ- 

 ences from the diagram of relationships based on 

 host morphology (Fig. 16). The only concordant sis- 

 ter groups produced in this tree are A canthocybium 

 and Scomberomorus. 



There are at least two problems with coding the in- 

 festation data in this manner. Use of copepod species 

 ignores information concerning the relationships of 

 the species. Another difficulty is coding copepod in- 

 festation as a two-state character (present or absent 

 in a host species), when Caligus infestation data can 

 only be interpreted as host preference (relative per- 

 cent of infestation) rather than as host specificity (see 

 previous section on Caligus). The program was rerun 

 using infestation by genera of copepods and defining 

 Caligus presence as more than 5% infestation to cor- 

 rect for this problem. This Wagner tree (Fig. 18) is 

 much closer to the diagram based on host morphol- 

 ogy. Several concordant sister groups are present: 

 Scomberomorus-Acanthocybium defined by the ac- 

 quisition of Tuxophorus at node (5), Grammator- 

 cynus- Scomberomorus + Acanthocybium defined by 



the acquisition of Caligus at node (4); Katsuwonus- 

 Thunnus, loss of Ceratocolax at node (9); and Eu- 

 thynnus-Katsuwonus + Thunnus, acquisition of 

 Pseudocycnus at node (8). 



There are also several differences between this 

 Wagner tree and the diagram of relationships based 

 on host morphology. Gymnosarda is associated with 

 Grammatorcynus-Acanthocybium group based on 

 the presence of Shiinoa in all four genera. However, 

 we found Shiinoa in only one specimen of Gymnosar- 

 da, so not much reliance can be placed on this asso- 

 ciation. We found only two other copepods on 

 Gymnosarda, single occurrences of C. bonito and C. 

 productus, which were omitted in this run of the pro- 

 gram. There was only one common copepod on Allo- 

 thunnus (Elytrophora), but there were also records of 

 the same two species of Caligus as in Gymnosarda. 

 Perhaps examining more specimens of Gymnosarda 

 and Allothunnus (we examined only seven of each) 

 would yield more copepods that would cluster these 

 two genera with the natural group of the Sardini 

 plus Thunnini. 



We turned from attempts at producing a cladistic 

 classification of all scombrids, using the infestation 

 data, and decided to use only a portion of the data, in- 

 festation by the nasal bomolochids of the genus Uni- 

 colax. The five known species of Unicolax are all 

 parasites in the nasal sinuses of scombrid fishes. The 

 first author compared characters within the species 

 of Unicolax with those in the related outgroup genus 



254 



