BREEDING TUBERCLES IN FISHES — COLLETTE 607 



First, they are present in the members of the tribe Etheostomatini, 

 subfamily Percinae, and in the members of the tribe Romanichthyini, 

 subfamily Luciopercinae. This fact might seem to suggest that the 

 Romanichthyini and the Etheostomatini should be placed together, 

 separate from the Percini and Luciopercini. When coupled with 

 other characters, such as the reduction of the air bladder, the small 

 body size, and the utilization of lotic rather than lentic habitats, this 

 arrangement seems even more reasonable. However, as I have 

 shown by using osteological characters (Collette, 1963), the relation- 

 ships of the Etheostomatini are with the Percini, and those of the 

 Romanichthyini are with the Luciopercini. Closer analysis of the 

 distributions of tubercles confirms this. In the Etheostomatini, 

 tubercles are concentrated on ventral and ventrolateral surfaces, 

 such as the undersides of the pelvic fins, the anal fin, the lower part 

 of the caudal fin, the scales on the belly, and the sides up to the 

 lateral line. In the Romanichthyini, tubercles are concentrated on 

 the dorsal and dorsolateral surfaces, the top of the head, the dorsal 

 fins, the pectoral fin, and the body scales prmiarily above the lateral 

 line. This difference in the distribution of tubercles may indicate a 

 different mode of spawning in the two tribes. Unfortunately, there 

 appears to be no information available about spawning in the members 

 of the Romanichthyini. Breeding tubercles and other characteristics 

 that are similar in the two tribes have probably arisen as independent 

 adaptations to a bottom-dwelling mode of life in streams. 



Within the Etheostomatini, patterns of tubercle development are 

 useful in showing phylogenetic relationships. The character has 

 no value at the generic level, at least if three large inclusive genera, 

 Percina, Ammocrypta, and Etheostoma, are recognized. However, 

 within each of the two larger genera (Percina and Etheostoma) tubercle 

 patterns offer an additional character in defining subgenera and species 

 groups within subgenera. 



In Percina, there are five nontuberculate subgenera {Hypohomus, 

 Alvoi'dius, Hadropterus, Sioainia, and Cottogaster) , two tuberculate 

 subgenera {Percina and Imostoma), and one heterogeneous subgenus 

 (Ericosma). On the basis of this character, perhaps the tuberculate 

 P. (Ericosma) evides is not as intunately related to the nontuberculate 

 P. crassa and P. palnnaris as previous workers have indicated. 



Males of all four species in the subgenus Ammocrypta and the one 

 species in the subgenus Crystallaria, genus Ammocrypta, are tubercu- 

 late. I think that the tubercle distributions in these species will be 

 found to be virtually identical, although I have not yet seen tubercles 

 on the anal fin of A. {A.) pellucida and I have found tubercles on the 

 caudal fin of A. {A.) clara alone. Thus, tubercle patterns agree with 

 other characters that lead to synonymizing Crystallaria with Ammo- 

 crypta, at least at the generic level. 



