POTTHOFF ET AL.: DEVELOPMENT OF SCOMBROID FISHES 



t* 



O) c 



*- 3 

 * 5 



_: * 



CD t- 



& > 



-" C 



CO 3 



TO 



bo <- 

 •C -° 



CO 



3 

 C 



s •- & 

 * p s 



bO <D 



cS cS 



IT) 3 J3 



ro O C 



* oj -O 



cS C 



_ "O si 



* %l 



3 

 0) 

 C 



0) 



CO CO 



c Si 



CS -j 



S s 



s -S 



■S.& 



8Q 

 ■a -9 



«- +J 



.O C 



CD CD 



t s 



cp a 



> o 



o % 

 a <o 



5 ^ 



"S "^ 



CO p 



CU fJh 



■rs !m 



o 



03 



£ 



OO 



I 



a. 



o 



I 



a> 



O CS 

 &H CS 



Thyrsites (Leionura), and Nealotus where only two 

 spines were supported (one supernumerary). Lar- 

 vae of Nealotus have three spines associated with 

 the first anal pterygiophore, but in juveniles the sec- 

 ond anal spine was fusing to the posterior process 

 of the pterygiophore. No evidence of a similar fu- 

 sion was observed in Rexea or Thyrsites (Leionura). 

 Gempylids had middle radials in one to six posterior- 

 most dorsal and anal pterygiophores (except Neo- 

 epinnula lacked middle radials). A double ray, and 

 a two-part posteriorly bifurcated stay was associated 

 with the last dorsal and anal pterygiophore in ap- 

 proximately one half of the genera. Lepidocybium, 

 Gempylus, Diplospinus, Paradiplospinus, Tongaich- 

 thys, and Rexea had a one-part posteriorly bifurcated 

 stay (Table 1). 



First caudal development of the cartilaginous 

 parhypural and hypurals 1 and 2 was concurrent 

 with anterior development of a few neural spines 

 and some first dorsal fin pterygiophores and fin 

 spines. The gempylid genera studied by us developed 

 all parts found in basic perciform caudal skeletons 

 (Gosline 1968), even the smaller second uroneural. 

 Caudal parts then fuse differently in the various 

 genera of adults (Matsubara and Iwai 1958). The 

 neural and haemal elements of preural centra 2 and 

 3 supported the procurrent caudal rays. In the gem- 

 pylids the procurrent spur on the posteriormost ven- 

 tral secondary caudal ray may be present, reduced, 

 or absent. Johnson (1975) examined two species in 

 which it was absent (Table 1). 



FAMILY TRICHIURIDAE 



Figures 5-8 



Seventy-three trichiurids in four genera were 

 available: 61 Trichiurus (4.5-26, 300, 303, 510 mm 

 TL), 8 Benthodesmus (4.5, 12 mm NL, 65-120, 541, 

 545 mm SL), 3 Evoxymetapon (210-550 mm SL), 1 

 Lepidopus (280 mm SL). Only Trichiurus yielded 

 a complete developmental series. 



Development of the vertebral column in Trichi- 

 urus initially started in two places on the noto- 

 chord: 1) anterodorsad (neural arch and spine on 

 future centrum 1), and 2) ventrad at the center (an- 

 terior haemal arches and posterior parapophyses). 

 Cartilaginous neural arches and spines were added 

 in a posterior direction. Haemal arches and spines 

 developed when the neural spines reached the ante- 

 rior future caudal vertebrae. Addition of haemal 

 arches and spines was also in a posterior direction 

 (Table 2). Trichiurus lacked a caudal complex. Ossi- 

 fication of the vertebral column started initially in 

 one place, with the anteriormost neural spines and 



659 



