GREGORY: FISH SKULLS 



319 



except in certain details in which it recalls the "Elopine" type; that the otoliths of the 

 trichiuroids are peculiarly specialized; those of the true Scombriformes are small, frail, 

 curved, and elongated, but "otherwise resemble the Elopine type." Since there is no 

 reasonable probability that the Scombriformes in the restricted sense have been derived, 

 independently of the carangids, from an elopine type, the evidence of the otoliths, in this, 

 as in certain other cases, does not in itself reveal the remote heritage of the group. 



The phylogenetic relations of the different branches of the scombriform fishes, as 

 inferred from the data reviewed above, are illustrated in Plate I. 



DiscocEPHALi (Sucking-disc Fishes) 



The amazing modification of an anterior spinous dorsal fin into a sucking-disc has 

 made these fish the objects of much study. But their real relationships within the Per- 

 comorphi remain a mystery. Dr. E. W. Gudger (1926) has called attention to the remark- 

 able resemblance between the young of certain echeneids and the young of Rachycentron, 

 a curiously modified percomorph. But a comparison of the skull, vertebral column and 

 fins of the adults lends no definite support to this view, as already noted by Boulenger 

 (1910, p. 691). The top of the skull (Fig. 198) has been considerably modified for the 

 support of the sucking-disc, notably by the widening of the mesethmoid and flattening of 

 the cranial table. The presence of parietals and the wide separation of the epiotics by the 

 supraoccipital and the lack of a "suborbital stay" (Fig. 199) indicate that this fish cannot 

 be derived from such a highly advanced percomorph as Scorpana, while the fairly normal 

 characters of the pectoral pterygials exclude the blennies and other advanced groups of 

 percomorphs. Comparison with the skull of the anacanth Lota reveals only general resem- 

 blances. The fairly normal predaceous mouth indicates a primitive percomorph ancestor. 



On the whole, these comparisons strengthen the inference that the echeneids have been 

 derived from relatively primitive percoids not dissimilar from the barrel-fish {Palinurich- 

 thys); this fish has the habit of lurking under floating logs and its arrector and depressor 

 muscles of the short spinous dorsal are unreduced. Once the habit of pressing the spinous 

 dorsal against the under side of the log was established as an indirect means of remaining 

 in the vicinity of a bountiful food supply, Natural Selection might soon be concentrated on 



PLATE I 



Inferred phylogenetic connections of the main branches of the scombriform fishes. From an exhibit in the American 

 Museum of Natural History. Drawings by Dudley Blakely chiefly from the monographs of Kishinouye. 



I. 1. Scomber japonicus 



2. Rastrelliger chrysosomus 



II. 1. Grammatorcynus bilinealus 



2. Sarda orientalis 



3. Gymnosarda nuda 



4. Thunnus thynnus 



5. Paralhunnus mebachi 



6. Germo germo 



7. Neothunnus macroptertis 



8. Auxis hira 



9. Euthynnus yaito 



10. Katiutoonus pelamis 



Key to Pedigree of Mackerels, etc. 

 III. 



1. Cybium Chinese 



2. Acanthocybium solandri 



IV. 1. Xiphias gladius 



2. Tetrapturuj imperator 



3. Isliophorui greyi 



V. 1. Ruveitus preliosus 



2. Epinnuta magiitralis 



3. Gempyla serpens 



4. Trichiurus lepturus 



5. Lepidopus caiidalus 



