FISHERY BULLETIN: VOL. 75, NO. 1 



The objectives of this study for each species 

 caught in sufficient abundance were: 1) to show 

 the relations between mesh size and the mean 

 length and standard deviation in length offish, 2) 

 to define gill net selectivity by applying the nor- 

 mal probability model, 3) to evaluate the applica- 

 bility of this model for defining selectivity, and 4) 

 to discuss uses of the derived information. 



STUDY AREA 



The study area was in the St. Andrew Bay sys- 

 tem located in northwest Florida along the Gulf of 

 Mexico. This bay system, compared to most other 

 northern gulf estuarine systems, is deep, has high 

 salinities, low freshwater inflows, large areas of 

 submerged marine grasses, low turbidities, high 

 percentages of sand in the substrate, and has fish 

 and crustacean faunas typical of both coastal and 

 estuarine areas (Ichiye and Jones 1961; Hopkins 

 1966; Brusher and Ogren 1976; May et al. 1976; 

 and Pristas and Trent 1977). The diurnal range of 

 the tide in the St. Andrew Bay system is about 

 0.5 m. 



ASSUMPTIONS 



The relation between the mesh size of gill nets 

 and the size of captured fish can be determined by 

 setting a series of gill nets that vary only in respect 

 to mesh size if certain precautions are taken and 

 certain assumptions are valid. Fishing effort must 

 be equal among mesh sizes, i.e., assume all fish of a 

 given length are equally likely to encounter all 

 nets. This means damage to each net must remain 

 low or about equal among mesh sizes, and net 

 locations are equal in respect to the probability of 

 a net catching a particular fish. We must assume 

 that no "gear saturation" occurs, i.e., the number 

 of fish already entangled in the net in no way 

 influences subsequent behavior of other fish and 

 the net, and that no "spill-over" occurs, i.e., large 

 fish do not lead along the nets until they encounter 

 a large enough mesh in which perhaps to become 

 enmeshed or entangled (Regier and Robson 1966). 

 We must further assume that loss offish from the 

 nets through predation is not dependent on mesh 

 size or the size of fish. 



GEAR AND METHODS 



Eleven gill nets, each of a different mesh size, 

 were fished for 126 days from 4 April to 29 De- 

 cember 1973 at a location about 400-1,000 m 



northwest of Courtney Point in St. Andrew Bay. 

 From 4 April through 20 September, the nets were 

 set every 14th day and fished for 72 consecutive 

 hours. From 20 September, the nets were fished 

 continuously until 13 December. The nets were set 

 again on 26 December and fished for 72 h. Nets 

 were anchored about 50 m apart parallel to each 

 other, perpendicular to shore, and in water depths 

 of 2.2 to 2.6 m (mean low tide). Nets were ran- 

 domized among net location each time the nets 

 were set. During the continuous fishing in the 

 autumn, the nets were randomized among lo- 

 cations twice during each 2-wk period. Net dam- 

 age to each net was maintained below 10% of the 

 total surface area. 



Increments of mesh sizes in the series of fished 

 nets were small, so that widely overlapping ranges 

 offish lengths would result. Mesh sizes used in this 

 study were chosen to catch the more abundant 

 species frequenting the St. Andrew Bay area 

 (Pristas and Trent 1977). Stretched-mesh sizes 

 ranged from 6.35 cm (2.5 inches) to 12.70 cm (5.0 

 inches) in 0.63-cm (0.25-inch) increments. 



The nets were 33.3 m long and 3.3 m deep. They 

 were made of #208 clear monofilament (0.33 mm 

 diameter, filament break strength about 26.4 kg) 

 nylon webbing. The webbing was hung to the float 

 and leadlines on the half basis (two lengths of 

 stretched webbing to one length of float or lead- 

 line, i.e., a hanging coefficient of 0.5). 



Fish were removed from the nets between 1 h 

 before and 2 h after sunrise and occasionally 

 between sunset and 1 h after. The total numbers of 

 each species, including damaged specimens, were 

 counted. Lengths of undamaged specimens were 

 measured to the nearest 0.5 cm. Fork length (tip of 

 snout to fork of tail) was measured for those fishes 

 having forked tails and total length (tip of snout 

 horizontally to extremity of the caudal fin) was 

 measured for Atlantic croaker, Micropogon 

 undulatus, and sharks. 



Length-frequency distributions of the catch by 

 species and mesh size, based on the number offish 

 that were measured, were adjusted to represent 

 the number of fish that were caught (those mea- 

 sured plus those damaged), so that the number 

 making up each distribution represented catch per 

 unit effort for each net. 



MODEL FOR 

 DETERMINING SELECTIVITY 



Basic mathematical models, or modifications of 



186 



