FISHERY BULLETIN: VOL. 75. NO. 1 



Tables 1, 2) for two of the smallest mesh sizes for 

 gulf menhaden and all mesh sizes for yellowfin 

 menhaden probably resulted from fish that were 

 caught by the opercles. This in turn probably 

 accounts for the curvilinearity of the data ob- 

 served for the two species of menhadens. A cubic 

 exponential equation such as that proposed by 

 Olsen ( 1959) might more accurately and precisely 

 define selectivity for gulf and yellowfin menhaden 

 over part of the length range of the selectivity 

 curve. 



The normal curve also provided acceptable 

 approximations to the data for sea catfish and 

 bluefish, although refinements in data collection 

 procedures, indicating how each fish was caught, 

 are needed to evaluate more accurately the model. 

 Sea catfish are frequently caught entangled by the 

 pectoral and dorsal spines, and bluefish are 

 frequently caught enmeshed or entangled by their 

 teeth, maxillaries, preopercles, and opercles. 



The normal curve did not provide acceptable 

 approximations to the data for pinfish and Spanish 

 mackerel. Pinfish were usually caught dorsally by 

 the dorsal antrorse spine and ventrally between a 

 point perpendicular to the antrorse spine and the 

 posterior end of the anal fin. With the fish and 

 webbing interacting in this fashion, the probabil- 

 ity of a given size of pinfish being caught was 

 probably about equal in a small range of mesh 

 sizes. The girth of a Spanish mackerel increases 

 gradually from its snout to the anterior point of its 

 second dorsal fin. Most individuals are caught 

 wedged in the mesh at any point between just 

 behind the opercle and the point of maximum 

 girth. The point of retention, therefore, is de- 

 pendent upon the mesh size within a small range 

 of mesh sizes. Also, many are entangled by the 

 teeth, maxillaries, and occasionally by the tail. 



Attempts to suggest models which might better 

 define selectivity for sea catfish, bluefish, pinfish, 

 and Spanish mackerel were not made in this 

 study, because the position at which each fish was 

 wedged in the net and — for those fish not wedged 

 in the net — the position at which each fish was 

 entangled was not recorded, and additional 

 catches of bluefish and Spanish mackerel were 

 needed. Holt (1963) suggested that, for species 

 that are caught at two or more distinct positions 

 along their body, selectivity could be defined by 

 regarding the selection curve as the algebraic sum 

 of two or more normal selection curves, or by 

 fitting an empirical curve such as the cubic ex- 

 ponential. Hamley and Regier (1973) found that 



the selectivity curve for walleyes was bimodal; 

 they resolved this curve into two unimodal 

 components representing fish that were caught by 

 wedging and entangling. 



Mean Length-Mesh Size Relation 



The second assumption of Holt's method is that 

 mean length of captured fish is proportional to 

 mesh size. To test this assumption, -2a/b was 

 plotted against the sum of mesh sizes (m,- + 1 + m,) 

 for each mesh-size pair (data from Table 2) and for 

 the seven species for which data for at least three 

 mesh-size pairs were available (Figure 2). Mean 

 selection length {alb or /,) in relation to mesh size 

 can also be determined from Figure 2 using the 

 bottom and right-hand scales. Data for Spanish 

 mackerel were plotted even though the assump- 

 tion of normality (previous section) for this species 

 was rejected. The straight lines in Figure 2 were 

 fitted through the origin by the least squares 

 method and the slopes (k) of these lines are given 

 in Table 2. With£ determined, the mean selection 

 length (/,-) for any mesh size is determined by /, = 

 m,k. 



Best fits of the data were obtained for Atlantic 

 croaker, blue runner, and yellowfin menhaden, 

 and acceptable fits were obtained for gulf menha- 

 den and sea catfish. More data are required, 

 however, to determine the degree of fit for the 

 remaining five species (bluefish, Spanish mac- 

 kerel, and the three species not shown in Figure 2). 

 Although the degree of fit cannot be evaluated for 

 the five species, information presented in Figure 2 

 or Table 2 can be used to provide rough estimates 

 of mean selection length in relation to mesh size 

 for bluefish, pinfish, spot, pigfish, and Spanish 

 mackerel. Much of the deviation about the re- 

 gression for bluefish (and possibly sea catfish) 

 probably resulted from fitting the line through the 

 origin (Figure 2). Apparently the mesh size-mean 

 length relation is not linear throughout a range of 

 mesh sizes between and 8.6 cm for bluefish. A 

 more reasonable approximation of the mean 

 length-mesh size relation for bluefish might result 

 by fitting a regular linear regression equation (Y 

 = a + bX rather than Y = bX) to the points in 

 Figure 2. For pinfish, spot, and pigfish, rough 

 approximations of the mean length-mesh size 

 relations can be obtained using the k value (Table 

 2 ) even though each k was based on only two points 

 and the origin. Variability about regression was 

 great for Spanish mackerel but this information 



192 



