data used in the analysis, the range of the data, and 

 whether sexes were analyzed separately. Consider- 

 ing the sample size (4) and the method of selecting 

 the points of overlap, the relationship for small blue 

 marlin (50-135 cm FL) was provisional. The rela- 

 tionship for shortbill spearfish was also provisional 

 since there were 16 data points ranging from 140.0 

 to 180.0 cm FL. Although the sample sizes for 

 black marlin, sailfish, and swordfish were small (24, 

 18, and 7, respectively), the ranges were wide, and 

 the relationships should be taken as valid estimates. 

 For striped marlin and for blue marlin, considering 

 all data sets, there were enough data to obtain valid 

 relationships. The importance of the results for the 

 various blue marlin data sets will be discussed in 

 connection with the coefficients of allometry. 



Concrete interpretations of the coefficients of al- 

 lometry are precluded by a statistical inability to 

 test the significance of all the coefficients as well as 

 to test between coefficients of different species or 

 data sets. The coefficient for swordfish was the 

 only one tested that was apparently greater than 

 3.0. For the other species tested, black marlin, blue 

 marlin (longline data), female blue marlin (trolling 

 data), sailfish, and shortbill spearfish, the 

 hypothesis that the coefficient was equal to 3.0 

 could not be rejected. That is, the growth in weight 

 to length was isometric for these species. Intui- 

 tively, we doubt these results for sailfish and short- 

 bill spearfish and suspect that additional data would 

 show the coefficient for sailfish to be less than 

 isometry and for shortbill spearfish to be greater 

 than isometry. 



For blue marlin, the interpretation of the results 

 was complicated by an inability to perform statisti- 

 cal tests of hypotheses. The coefficient of allometry 

 for the small blue marlin indicated that the small 

 fish maintain a very different weight to length 

 growth relationship than do the larger, adult fish. 

 Part of this difference may have been due to differ- 

 ential growth of the bill in the younger fish. It was 

 apparent from Table 4 that there was not a real 

 difference between longline- and troll-caught blue 

 marlin; the coefficients of allometry as well as the 

 intercept "b" were extremely similar. This does 

 not necessarily imply that there are no seasonal dif- 

 ferences in the weight-length relationship of blue 

 marlin but does indicate that no such effect could be 

 shown with 68 data points from longline catches 

 made over all seasons. When the trolling data were 

 divided according to sex, it was found that the coef- 

 ficient for females did not differ significantly from 



BLUE MARLIN 



i T — r - 



i i , 



l I 



i 



50-135 CM FL „„„ 

 W= 0.5183 FL a667B 











£135 CM FL 



W= 4.7226 X io" 6 FL 3 - 0442 











N = 68 









- 



" 





'/* 





- 



~ 









- 





• 5^ % 







- 



• — i 1 r**^i i 













MALES AND FEMALES 



' 



1 1 1 T 1 







. 



300 



W = 4.2968 X iff" 6 Ft 10470 

 N - 384 





* / 



. 



200 



" 







- 





- 



it 





- 



100 





WT' 





- 



O 



1 1 1 1 1 1 1 L 





.lit 



i 



FEMALES 



W= 1.9445 X I0" 6 FL 31871 

 N = 85 



a 2oo 



-i 1 _l L_ 



_J I L_ 



MALES 

 w = 2.2929 X I0* 5 FL 27405 



250 



300 



350 400 



LENGTH (CM) 



Figure 5. — Weight-length relationships using the log- 

 linear model for blue marlin. The upper chart represents 

 the relationships found for small and large fish using long- 

 line data. The remaining three charts represent relation- 

 ships for sexes combined (including sex undetermined), 

 females, and males using trolling data. The aberrant 

 datum appearing in the sexes combined and female charts 

 for the trolling data was not used in the calculation of the 

 relationships. 



isometry while that for males was probably less 

 than isometry. The male and female curves (Fig. 5) 

 could not be distinguished where the data over- 

 lapped. Hence, the increased weight to length 

 growth shown by the females occurs primarily at 

 lengths greater than those attained by males in this 



136 



