FISHERY BULLETIN: VOL. 85, NO. 4 



Table 7. — Results of analysis of covariance for growth differences observed in 

 Spanish mackerel collected in northwest Florida, and fish collected in all areas 

 by recreational hook and line, and gill net. 



(sex, area x sex, and age were also included in the 

 covariance model) for recreational hook and line 

 samples and gill net samples. Area differences 

 were highly significant for both gear types, and 

 sex differences were highly significant for gill 

 net-caught fish, but somewhat less so for hook 

 and line samples (Table 7). The area x sex inter- 

 action was significant for hook and line, but not 

 gill net samples. 



These ANCOVA results demonstrate that fe- 

 males grew significantly faster than males. The 

 significant differences between sampling gears 

 are no doubt due to gear selectivity, i.e., hook and 

 line selecting for larger fish of a given age and gill 

 nets selecting for a specific size fish. Significant 

 differences between sampling areas (consistent 

 for both sampling gears) substantiate faster 

 growth in south Florida (fish were larger at a 

 given age) than in northwest Florida or Louisi- 

 ana. 



We compared back-calculated lengths-at-age of 

 Spanish mackerel (from all areas and from Flor- 

 ida alone) with those of Powell (1975); lengths at 

 ages 1 and 2 for both sexes were shorter, while 

 those for ages 3-5 were increasingly longer (Table 

 8). There was a greater discrepancy between our 

 data and Powell's for males than for females. 

 Florida males from our study were 38 mm shorter 

 than Powell's at age 1, but by age 5 they were 120 

 mm longer. Florida females from our study were 



Table 8. — Mean back-calculated fork lengths (mm) at age by sex 

 for Spanish mackerel from Powell (1975) and this study. Powell's 

 data were transformed from standard length by his formula 

 FL= 1.0728 SL + 2.4267. 



25 mm shorter than Powell's at age 1, but by 

 age 5 they were 33 mm longer. Some of this 

 discrepancy can be explained by the fact that 

 Powell used the direct proportion method for his 

 back-calculations, whereas the program by 

 Abramson (1971) employs the regression method. 

 Carlander (1981) pointed out potential problems 

 with this method, but they primarily concern the 

 fact that when using the scales for ageing, not all 

 scales on a fish are the same size. This problem is 

 of lesser importance when ageing is done from 

 otoliths. 



Our estimates of the von Bertalanffy growth 

 coefficient (k) are smaller, and our asymptotic 



782 



