(2) for facility in measuring or counting, or 



(3) as a conscious selection of characters likely 

 to show possible differences . 



Schaefer (1948), in his initial paper on 

 morphometnc characteristics of Pacific yellow - 

 fin tunas, reported on moiphometric data for 

 46 yellowfin tuna from the waters off Costa Rica. 

 He gives data on 21 of the 27 characters defined 

 by Marr and Schaefer. For each character, 

 Schaefer determined the linear least-squares 

 regression on the total length (or head length) . 

 For most characters studied, the regression of 

 the character on total length was a rectilinear 

 regression of the type Y = a + bx. For three 

 characters, it was necessary to transform the 

 original data (using logarithms for one or both 

 of the measurements) in order to yield a recti- 

 linear relationship. The rate of increase in the 

 length of second dorsal and anal fins was found 

 to be greater than that of total length, while the 

 rate of growth of the pectoral fin was less than 

 that of total length . 



Schaefer and Walford (1950) compared 

 yellowfin tunas of Angola and the Pacific Coast 

 of Central America . Morphometric data on 

 Angola yellowfin included all 27 characters de- 

 fined by Marr and Schaefer. The regression of 

 body parts on standard length yielded results 

 very similar to that found for Costa Rican tuna . 

 All characters had a linear relationship with 

 total length except length of second dorsal, anal 

 and pectoral fins . 



The most striking difference exhibited by 

 tunas of the two areas was in the length of the 

 pectoral fin relative to the length of the body . 

 In both regions the length of the pectoral fin 

 relative to total length decreased as the size of 

 the fish increased, but the rate of decrease was 

 more rapid for Costa Rican yellowfin . Similarly 

 marked differences were found between the rela- 

 tive growth of the second dorsal and anal fins. 

 In both proportions the Atlantic specimens had 

 relatively larger fins than the Pacific specimens. 

 Regression of the following characters on total 

 length was slightly but quite significantly differ- 

 ent between the two regions: head length, dis- 

 tance from snout to first dorsal, from snout to 

 second dorsal and from snout to anal fin. No 

 significant differences were found in the other 

 characters studied. 



The technique used in comparing the re- 

 gression lines of tuna from two regions is that 

 of covariance analysis. It provides a precise 

 answer to the question of whether two samples 

 differ more than would be expected from chance 

 variation in samples from the same population. 

 If such a difference exists it is good evidence of 

 the heterogeneity of the population. It is not 

 evidence that the stocks are completely separate 

 (although they may be), nor does the technique 

 afford a means of assessing the extent to which 

 stocks may intermingle. However, such tech- 

 niques are described in the papers in this series 

 by Royce and by Widrig and Taft . 



Schaefer (1952) extended his study to the 

 yellowfin tunas of Hawaiian waters. He found 

 nearly as marked differences between Hawaiian 

 and Costa Rican yellowfin as he had found between 

 Angolan and Costa Rican specimens . In fact, in 

 the relative length of the dorsal and anal fins he 

 found Hawaiian yellowfin to differ more markedly 

 from Costa Rican specimens than the Angolan 

 specimens had differed from the Costa Rican. 



A count of gill rakers on 188 Hawaiian tuna 

 had a mean value of 29.66 with a standard error 

 of 0.087. Counts on yellowfin of the American 

 West Coast averaged 30.46 with a standard error 

 of 0.116 for 115 specimens. There is no correla- 

 tion between size of fish and gill raker count, 

 hence these characters avoid the difficulties of 

 regression analysis. Consequently, this char- 

 acter seems to offer good possibilities for racial 

 analysis of yellowfin tuna (ibid : 371) . 



The regression coefficients of various 

 dimensions on total length for the three regions 

 are given in the table on the following page . 



In Hawaiian material, head length and the 

 distance from the snout to the various fin inser- 

 tions increase more slowly, relative to toal length, 

 than in Costa Rican material, while the caudal 

 portion of the body grow^ fastef • As a result, 

 large specimens of Hawaiian yellowfi n are more 

 elongate in the posterior portion of the body than 

 specimens of comparable size from Costa Rica . 



Schaefer concludes: There is no doubt 

 that the two populations (Hawaii vs. Costa Rica) 

 are to be regarded as distinct. The possibility 

 of some mixing between them is not excluded, 



59 



