﻿REVIEW OF GENUS DOROSOMA — MILLER 389 



the body began with the first scale just in front of the base of the left 

 pelvic fin, counting upward in zigzag fashion over the back and down 

 the right side over the belly to the starting point. The scale number 

 around the peduncle was also made in zigzag fashion around the 

 slenderest part. 



The count of ventral scutes was divided into two parts, prepelvic 

 and postpclvic. The tip of the last prepelvic scute counted extended 

 to or close to the insertion of the pelvics. Thus my counts are in 

 agreement with those of most workers except Hubbs and Miller 

 (1941), who regarded the last prepelvic scute as always extending 

 bej^ond the pelvic insertions. Their prepelvic counts should there- 

 fore be decreased by one and their postpelvic counts increased by one 

 (e. g., 19+11 = 18+12) to agree with general practice. In enumerat- 

 ing the vertebrae, the hypural was always included. Whenever any 

 count was in doubt, the maximum possible figure was always recorded. 



In measuring head length I always included the opercular mem- 

 brane. Head depth was measured vertically over the posterior end 

 of the isthmus. The dorsal filament was measured from its tip to the 

 posterior face of its contact with the back. 



THE GENERA DOROSOMA AND SIGNALOSA 



Jordan and Evermann (1896-1898, pp. 415, 2809-2810) and others 

 placed the closely related genera Dorosoma and Signalosa in a separate 

 family, the Dorosomidae (more properly, the Dorosomatidae). Later 

 students, adoptmg a more conservative view, generally have referred 

 these genera and their Old World relatives, Nematalosa, Gonialosa, 

 Anodontostoma, Konosirus, and Clupanodon, to the Clupeidae.^ The 

 describers of Signalosa, Evermann and Kendall (1898, p. 127), also 

 referred that genus to the Clupeidae, and in the most recent general 

 classification of fishes by Berg (1940, p. 421) Dorosoma and its allies 

 are given subfamily recognition only. 



The generic status of Dorosoma has been reviewed recently in detail 

 by Hubbs and Miller (1941, pp. 233-234). In addition to structural 

 characters that differentiate Dorosoma and Signalosa, I now find that 

 Signalosa has only 40 to 45 vertebrae, in contrast to 43 to 51 in 

 Dorosoma (table 1). In the areas where the two genera are known to 

 occur together, from Florida to northern Guatemala, the vertebral 

 counts do not overlap. The fewer vertebrae were briefly noted by 

 Regan (1917, p. 310) who gave 41 as the number in Signalosa. Thus 

 the New World Dorosomatinae have 40 to 51 vertebrae, rather than 

 49 as stated by Jordan and Evermann (1896, p. 415). 



Despite the review of Signalosa by Weed (1925), the members of 

 this genus are still in need of clarification, as Gunter (1945, p. 31) has 



1 Among recent writers, Fowler (1945, p. 22) retained family recognition (Dorosomidae) for Dorosoma and 

 Signalosa. 



