CRESSEY ET AL.: COPEPODS AND SCOMBRID FISHES 



not concordant. We also know that several parts of 

 these data sets are in agreement, that is to say, some 

 evolutionary events in Unicolax are correlated with 

 speciation (vicariant) events in the Scombridae. 

 These events are easily explained by models of allo- 

 patric speciation and hypotheses of dispersal are un- 

 necessary. Before we can suggest a dispersal event, 

 we must first factor out host-parasite relationships 

 which are due to cospeciation events. They may be 

 done by overlaying parasitic phylogenetic data in the 

 form of a character state tree on the host phylogeny. 

 This procedure is similar to the generation of the host 

 by parasite tree (Fig. 24), with the exception that the 

 parasite phylogenetic information is forced onto the 

 host cladogram. 



In our example the scombrid host tree was coded as 

 a character state tree. A character by scombrid taxon 

 matrix was constructed so that each character was 

 repeated a number of times. To this we added the 

 characters from the parasite phylogeny by host data 

 matrix used to generate the host by parasite tree. The 

 repetition of the character by scombrid taxa matrix 

 has the effect of forcing the tree into a particular 

 shape, in our case, the original host cladogram. The 

 number of replicates is large enough so the parasite 

 phylogeny data does not alter the outcome of the 

 tree. This combined data matrix was submitted to 

 the WAGNER 78 program and a most parsimonious 

 tree was generated. This tree (Fig. 26) is the same 

 shape as the original host phylogeny, and characters 

 relating to historical events of the parasites are over- 

 layed or forced onto the tree in a parsimonious con- 

 figuration. 



The overlay presented in Figure 26 indicates that 

 parasite evolutionary events (— 2), (17), (— 8), and 

 (—9) (indicated as characters circled in broken lines) 

 were reversed or lost in several host taxa or lineages. 

 This indicates the loss of a parasite or a hypothetical 

 ancestral parasite. The only independent acquisition 

 of parasites or hypothetical ancestral parasites oc- 

 curred between Sarda, node (8), and Auxis on the 

 cladogram. In both cases parasite 4, U. mycterobius, 

 and its hypothetical ancestors (6) and (7) not only 

 were independently acquired but also must have 

 been independently evolved. In this case it is more 

 reasonable to invoke an hypothesis of dispersal and 

 to explain the infestation of Sarda by U. mycterobius 

 by dispersal from another scombrid host. This hypo- 

 thesis is more parsimonious than the coevolutionary 

 hypothesis in that it requires one dispersal event 

 rather than a series of independent identical evolu- 

 tionary events (having serious taxonomic im- 

 plications for parasitic taxa, i.e., if two taxa evolve in 

 independent lineages they must be considered sepa- 



CO 



a 

 c 



c 



3 



•x. 



3 



CO 



3 

 C 

 C 

 iv. 

 ■C 



3 

 Uj 



CO 



3 

 C 



O 



5 



CD 



CO 



3 



c 



3 

 •C 

 I— 



1 



Kr 



M4* 



(&* 



U 



11 



J 



7 = 



J 



J 



J 



I 



Figure 26. — Overlay of historical parasite information on host 

 phylogeny. Negative numbers indicate losses and numbers circled 

 in broken lines indicate independent acquisitions or losses of para- 

 sites or parasite ancestors. 



rate, possibly sibling species). It must be noted that 

 an hypothesis of independent evolutionary events 

 leading to the establishment of U. mycterobius on 

 Sarda may in itself require a dispersal event earlier in 

 its evolutionary history. 



The coevolution of Unicolax and its scombrid hosts 

 can be reconstructed as follows. The three higher 

 tribes of the Scombrinae (Scomberomorini, Sardini, 

 and Thunnini) share Unicolax, indicating that this 

 genus arose from a more primitive bomolochid after 

 the ancestors of these three tribes evolved from the 

 Scombrini. Unicolax ciliatus, the most primitive 

 species of Unicolax, is present only in the most primi- 

 tive of the three tribes, the Scomberomorini. Uni- 

 colax collateralis is found on members of the tribes 

 Sardini and Thunnini. Infestation by U anonymous 

 yields little information because it is restricted to Eu- 

 thynnus alletteratus from both sides of the Atlantic. It 

 is apparently a more recently derived species that 

 has not spread far geographically or host-wise. Un- 

 icolax mycterobius is restricted to the two most primi- 

 tive genera of the Thunnini (Auxis and Euthynnus) 

 except for its presence on two specimens of Sarda 

 orientalis from Japan. This seems best explained as 



259 



