FISHERY BULLETIN: VOL. 75, NO. 1 



sizes used and were usually caught entangled by 

 their snout and caudal fin; they were rarely 

 wedged in the meshes. Bonnethead sharks were 

 almost always caught in meshes that had been cut 

 (probably by the sharks) and with their teeth 

 entangled in adjacent meshes; because of these 

 circumstances we did not expect a correlation 

 between the size of shark and mesh size. 



Based on the data requirements of Holt's 

 method, only the 10 most abundant species (Table 

 1) were selected to evaluate one or more of the 

 three assumptions — normality of selection curve, 

 linearity of mean length-mesh size relation, and 

 constancy of standard deviation between mesh 

 sizes — required for Holt's model. For these species, 

 length-frequency distributions for those mesh 

 sizes where n ( >50 are shown in Appendix Tables 

 1-3. These distributions are provided as the basis 

 for our evaluation of selectivity and for applying 

 other mathematical models to the data if other 

 investigators so desire. 



SPECIES CAUGHT IN 

 GREATEST ABUNDANCE 



Normality of Selection Curves 



Natural logarithms of the ratios (lnR, + 1(/ ) of 

 numbers offish of length /, caught in meshes m ( + 1 

 and rm were plotted against lengths of fishes to test 

 normality of the selection curves. Least squares 

 regression equations were computed, and the 

 intercepts (a) and slopes (b) of these equations are 

 shown in Table 2. 



Best fits of the points to the straight lines were 

 obtained for spot, Leiostomus xanthurus; pigfish, 

 Orthopristis chrysoptera; Atlantic croaker; and 

 blue runner. The mean values of s vx [standard 

 deviation of Y (ratio) for fixedX (length) in linear 

 regression analysis (Steel and Torrie I960)] were 

 lowest for these four species and ranged from 0.211 

 to 0.319 (Table 2). Slight curvilinearity appeared, 

 however, in the data for the 7.0/6.3 and 7.6/7.0 cm 



TABLE 2. — Coefficients of, and estimates from, least squares regression equations of lnR +1  on 

 length by species and mesh-size pair, and k values by species. 



190 



