NO. 3 



TELEOSTEAN FISHES — GOSLINE 



37 



In conclusion it seems relevant to discuss groupings of modern 

 teleostean orders. The older divisions such as that between the Mala- 

 copterygii and Acanthopterygii or between Physostomi and Physo- 

 clisti need no present consideration. The defect of all such systems 

 lies in the fact that any one character may have been lost or gained 

 independently many times in teleostean evolution. That independent 

 lineages may lose a character, e.g., the connection between the air 

 bladder and the esophagus, is fairly obvious. That a character may 

 be and often is gained in independent lineages is sufficiently indicated 

 by the fact that the consolidation of lepidotrichia into spinelike struc- 

 tures has taken place time and again in teleosts, e.g., in the Cyprini- 

 formes, Notacanthi formes, Cyprinodontiformes, and Perciformes. As 

 with the protrusile upper jaw, teleosts seem to have tried out fin spines 

 in a number of ways before developing the generally satisfactory 



Orders with the following percoid-type structures : 



Branchiostegal rays 



Syngnathiformes 



Percopsiformes 



Cyprinodontiformes 



Gadiformes 



Macruriformes 



Lampridiformes 



Gasterosteiformes 



Beryciformes 



Zeiformes 



Higher orders 



Protrusile premaxillaries Pelvic spine 



Gadiformes 



Macruriformes 



Lampridiformes (?) 



Gasterosteiformes 



Beryciformes 



Zeiformes 



Higher orders 



Gasterosteiformes 

 Beryciformes 

 Zeiformes 

 Higher orders 



Diagram 3. 



type found in the perciform fishes; or, alternatively (as suggested by 

 the notacanthids), a generally adaptable type of fin spine may have 

 been developed only after the lineage had specialized too far in other 

 ways to allow much adaptive radiation. In any event, a formal divi- 

 sion of the teleostean fishes based on a single character, whether one 

 of loss, e.g., Physostomi, or of new development, e.g., Acanthopter- 

 ygii, is open at least to grave suspicion. 



Insofar as such a basis of division is valid at all, the characters of 

 the hyoid arch (Hubbs, 1919), the development of a percoid type 

 protrusile upper jaw, and the formation of a single "true" outer 

 pelvic spine would seem to provide the most useful dividing lines 

 (diagram 3). As may be seen from this diagram these three features 

 furnish a division between a "lower" and a "higher" group of orders 

 at roughly but not exactly the same point. 



