more dorsally situated ; by the 20th to 25th vertebra 

 it comes to arise from the anterior end of the centrum 

 rather tlian from the haemal arch. 



Beginning at about the fourth vertebra, a process 

 which has been called a haemal postzygapophysis 

 (de Sylva, 1955) arises on each side from the poster- 

 ior end of each centrum. On the anterior centra it is 

 small and laterally directed. This process becomes 

 ventrally directed on the eighth vertebra, and its 

 distal end meets the parapophysis ; farther poster- 

 iorly it meets the upper part of the haemal arch or 

 the haemal prezygapophysis when it is formed. On 

 approximately the last eight vertebrae the haemal 

 spines are situated so posteriorly that they obliterate 

 the haemal postzygapophyses. 



From some or all of the 20th to the 33d vertebrae, 

 the blood vessels and nerves that emerge from the 

 haemal canal exit through ventrolateral foramina. 

 The anterior foramina are formed by struts running 

 from the haemal arch to the centrum near the base 

 of the haemal postzygapophyses. They become 

 smaller posteriorly and are separate from and ante- 

 rior to the haemal arches; the latter in this region 

 gradually become located toward the posterior end 

 of the centra. On the 32d to 36th vertebrae, flat- 

 tened lateral processes form a horizontal bony keel. 

 The sizes of the individual vertebrae vary consider- 

 ably, and regional differences are emphasized in 

 older specimens. The length increases regularly to 

 the 35th vertebra; the 36th is slightly shorter, the 

 37th and 38th are very short, and the 39th is 

 incorporated into the wide, triangular hypural plate. 

 The depth of the vertebra increases regularly to 

 about the 25th, beyond which there is a gradual 

 decrease to the hypural plate. A simple splintlike 

 hone (epural) is closely applied to the anterodorsal 

 surface of the hypural plate. A similar bone 

 (hypural) bearing a spinous process on each side is 

 present along the anteroventral surface of the 

 hypural plate. The terminology and derivation of 

 these two bones are in doubt. 



Specific Characters. — The vertebrae typically total 

 39 in all species. Godsil and Byers (1944) reported, 

 and we have reexamined, a California T. thyniius 

 with only 38, in which 1 vertebra near the hypural 

 is obviously missing. Among more than 200 skele- 

 tons of the seven species, we found only three 

 additional abnormalities, all due to recognizable 

 fusion of two adjacent centra. Frade (1932) re- 

 ported, among 110 T. thynnus, 8 with 38 vertebrae, 

 6 with 40, and 1 with 41. We doubt the counts of 



40 and 41 but cannot explain them. All but one 

 species have 18 precaudals and 21 caudals, the first 

 long haemal spine occurring on the 19th vertebra. 

 The same count was given by de Sylva (1955) for 

 T. atla)}ticus, but, as Watson (1964) has shown, this 

 species differs from all other Thunnus in having 19 

 precaudals and 20 caudals. Exceptions may be 

 expected in all species; we have examined T. atlan- 

 ticus with counts of 18-|-21 and 20+19, T. obesus 

 and T. albacarcs with 17 + 22, and T. thynnus and 

 T. obesus with 19 + 20; and Godsil and Byers (1944: 

 86) reported one T. alalunga with 20+19. 



The position of the first (anteriormost) ventrally 

 directed parapophyses appears to show almost no 

 variability within a species or subspecies; only one 

 exception has been noted. These parapophyses 

 occur on the 8th vertebra in T. thynnus, on the 10th 

 in T. tonggol, and on the 9th in the other species. 

 In T. alalunga none of the parapophyses is directed 

 quite so obviously ventrad as in the other species; 

 those on the ninth vertebra that we regarded as 

 ventrally directed are much shorter than in any 

 other species and seem almost twisted, never becom- 

 ing completely ventrally oriented. In other species, 

 there is variation in the ventral extent of the preced- 

 ing parapophyses. As long as these were more or 

 less flattened and rounded, their relative location 

 was not considered. The first ventrally directed 

 ones are definitely elongated in a ventral direction 

 (compare the eighth vertebra in T. thynnus and 

 T. niaccoyii, fig. 10). 



The first (closed) haemal arch usually occurs on 

 the 11th vertebra in T. albacarcs. T. atlanticus, T. 

 tonggol, and T. obesus, and usually on the 10th 

 vertebra in T. alalunga, T. maccoyii, and T. thynnus. 

 In all species except T. alalunga and T. maccoyii we 

 observed the first closed arch occasionally either one 

 vertebra anterior or one vertebra posterior to the 

 usual position. Godsil and Byers (1944: 68, 101) 

 observed notable variation in Pacific T. albacarcs 

 and in T. thynnus. In most of the species the 

 parapophyses on the vertebra preceding the one that 

 bears the first haemal arch approach each other so 

 closely in the median line that it appears to be a 

 matter of chance whether or not they or the next 

 pair fuse. In many specimens in which the haemal 

 arch was formed anterior to its usual position, its 

 shape was noticeably different (fig. 10, OBE). 



In T. alalunga the first haemal arch is directed 

 forward, Avith an angle of about 45° or less between 

 it and the vertebral axis. In all of the other species 



ANATOMY AND SYSTEMATICS OF TUNAS 



77 



