Regression coefficients of various dimensions on total length 



Character 



Head length 

 Snout to 1st dorsal 



" " 2nd 

 " anal 



" " ventrals 



Length pectorals (on log TL) 

 Log length 2nd dorsal (on log TL) 

 " " anal (on log TL) 



but if any exists it must be sufficiently small to 

 permit the two populations to maintain their 

 characteristic differences" (ibid. : 3 72) . 



Godsil also has done considerably work 

 on "populations" of yellowfin tuna (Godsil 1948, 

 Godsil and Greenhood 1951). He has dealt with 

 fewer characters than Schaefer (five in addition 

 to body length), but has examined a much larger 

 amount of material. He studied 13 samples 

 containing a total of 1,911 fish from the area 

 regularly fished by the American tuna fleet 

 (from Gorda Bank, Mexico to the Galapagos Is.). 



Godsil decided that the regressions of 

 the characters studied (head length and distance 

 from snout to fin insertions) on total (fork)length 

 were nearly but not quite linear . He got the 

 best fit with the function Y = a + bx + c/x. 



Godsil tested the homogeneity of his 13 

 "American" samples and found that they could 

 not be considered a homogeneous population . 

 He then tried to find if the heterogeneity was 

 associated with geographical origin. If such 

 were the case, duplicate samples from the same 

 region should be homogeneous, and it should be 

 possible to obtain homogeneity by a proper and 

 natural grouping according to origin. 



Three of his samples were obtained from 

 the same locality off Costa Rica. This is an in- 

 teresting group, for they were caught from the 

 same locality that Schaefer obtained his Costa 

 Rican specimens. The regression coefficients 

 for regression of the five dimensions on total 

 length are given for Costa Rican samples in 

 table 6. On testing for homogeneity Godsil found 



that the regressions of three of the five dimen- 

 sions were significantly different between samples . 

 Samples obtained from the same locality, conse- 

 quently, were found to differ as much as samples 

 from widely separated regions, hence it was con- 

 cluded that the. heterogeneity of the data is due to 

 causes other than geographic segregation. 



Godsil and Greenhood (1951) have also 

 studied "populations" of yellowfin tuna in the 

 Central Pacific . They had measurements on 79 

 yellowfin from Hawaiian waters, 94 from off 

 Palmyra and Fanning Islands (4-6° N. lat., 159- 

 162° W. long. ,) and 13 from off the Fiji Islands. 

 As in the earlier paper, five characters were 

 used: head length, and the distance from snout to 

 fin insertion (1st D, 2nd D, anal and ventrals). 



They found that the Central Pacific yellow - 

 fin were significantly different from eastern 

 Pacific yellowfin in all characters studied. The 

 regression lines from central Pacific stocks were 

 below the regression lines for eastern Pacific 

 yellowfin for all characters. They concluded that 

 the central Pacific stocks of yellowfin were dis- 

 tinct from and non -intermingling with those of the 

 eastern Pacific . There can be little doubt that 

 the two stocks are not homogeneous, but the 

 authors have not demonstrated that intermingling 

 does not take place . 



Royce (1952) has summarized information 

 on stocks of yellowfin tuna in the Pacific. In ad- 

 dition to the data already discussed, he included 

 morphometric measurements from a number of 

 other areas including the waters around the Line 



60 



