FISHERY BULLETIN: VOL. 79, NO. 1 



nation of the appropriateness of the basic catch 

 model for a single species-area case. Separate 

 analyses were performed for blue marlin, white 

 marlin, and sailfish in each of the three recre- 

 ational areas. 



Since the model assumes that the catchability 

 coefficients of dock and tournament fishing are 

 proportional [Equation (3)], correlation analysis 

 was performed on the dock and tournament CPUE 

 values (Phase 1, Figure 7) and the model was con- 

 sidered appropriate for estimating fishing power 

 only when the CPUE's were significantly corre- 

 lated at the 5% level. Data used in the correlation 

 analysis were from all months in which dock and 

 tournament fishing met the minimum effort 

 threshold concurrently. 



The two factor ANOVA model [Equation (2)] was 

 then used to test for significant differences in fish- 

 ing power and density, and Tukey's (1949) test was 

 used to test for significant interaction. The data 

 used in the ANOVA were from all months in the 

 1971-78 period for which dock and tournament 

 sampling met the minimum effort threshold con- 

 currently and for which CPUE's were >0 for both 

 types of fishing. The positive CPUE constraint was 

 necessary because of the log transformation used 

 in obtaining Equation (2). Because the model re- 

 quires that there be no interaction between power 

 and density, the model was not considered appro- 

 priate when interaction was significant at the 5% 

 level. 



For all cases in which the model was deemed 

 appropriate and the ANOVA test for difference in 

 power was not significant at the 5% level, the catch 

 and effort data were pooled and a single recre- 

 ational CPUE was calculated for those species- 

 area combinations. Where the model was appro- 

 priate and the power was significantly different, 

 dock sampling was designated as the standard and 

 the relative fishing power of tournament fishing 

 was estimated from Equation (5). The computer 

 program FPOW (Berude and Abramson 1972) was 

 used to estimate the relative fishing power. FPOW 

 solves the normal equations like Equation (5) and 

 corrects for the logarithmic bias using a Taylor 

 series expansion of the estimate about its true 

 value (Laurent 1963). The FPOW program was 

 modified to perform the usual F-test for the sig- 

 nificance of the overall regression and to compute 

 the coefficient of determination. As in the ANOVA 

 test, the data used in the fishing power estimation 

 were from all months for which dock and tourna- 

 ment fishing met the minimum effort threshold 



concurrently, and for which both CPUE's were >0. 

 For those species-area combinations in which the 

 model adequately represented the recreational 

 data, the entire procedure was then repeated in an 

 analogous manner to compare the recreational 

 and longline data. 



Results 



The results of the correlation, ANOVA, and re- 

 gression analyses for blue marlin, white marlin, 

 and sailfish from the Panhandle, New Orleans, 

 and Texas areas are summarized in Table 5. 



Blue Marlin 



In the Panhandle area, dock and tournament 

 CPUE data are fairly consistent and it appears 

 that fishing power is greater for dock data than for 

 tournament data. When the dock and tournament 

 data were pooled and compared with the longline 

 data, no correlation was found and interaction 

 between power and density was apparent. In the 

 New Orleans and Texas areas, no significant dif- 

 ference in the power of dock and tournament data 

 was found, but the CPUE's were not correlated 

 and interaction was significant in the New Or- 

 leans data. 



The blue marlin results generally indicate that 

 the basic catch model does not adequately repre- 

 sent the blue marlin data in the northern Gulf of 

 Mexico. While it may be possible to obtain 

 adequate indices of abundance from recreational 

 or longline data, the two types of fishing appear to 

 be providing very different indices in the same 

 local areas, and it cannot be determined which, if 

 either, provides a valid measure of relative abun- 

 dance. It appears that until the dynamics of the 

 blue marlin fishery are better understood, the use 

 of nominal catch and effort data to index relative 

 abundance may produce inconsistent and mislead- 

 ing results. 



White Marlin 



In the Panhandle and New Orleans areas, the 

 CPUE's were well correlated, no significant differ- 

 ence in the power of dock and tournament data 

 was found, and no interaction was apparent. When 

 dock and tournament data were pooled and com- 

 pared with longline data, the CPUE's were well 

 correlated, and a significant difference in power 

 was found, but significant interaction was found in 



60 



