252 



Fishery Bulletin 94(2), 1996 



were used to compare prey composition and prey size 

 between bagged and unbagged fish. All tests were 

 two-tailed (a=0.05). The Bonferroni correction 

 (a'=0.05/m; Manly, 1991, p. 52) was used to control 

 type-I error for m = 3, 2 x 2 chi-squared tests of pres- 

 ence or absence of prey types between treatments. 

 Data were too few to analyze by separate chi-squared 

 tests for each date. A three-way G-test was inappro- 

 priate for simultaneous comparisons of prey types be- 

 cause the occurrences of types were interdependent. 



or elsewhere in the pharynx of both bagged and 

 unbagged fish. Divers witnessed unbagged snapper 

 frequently regurgitating stomach contents between 

 15-18 m depth and the sea surface, as the volume of 

 gases in swimbladders expanded by more than an 

 atmosphere of pressure, equivalent to 40°< of all blad- 

 der expansion between depths of 75 and 15 m. Snap- 

 per were never seen regurgitating prey below 18 m, 

 although safety considerations precluded prolonged 

 observations by divers at depths >18 m. 



Results 



Fish specimens 



A total of 180 juvenile snapper (88 bagged, 92 

 unbagged) were collected on the nine sampling dates. 

 From 5 to 40 specimens were collected per date (num- 

 bers of bagged and unbagged fish were <15% differ- 

 ent on six of the last seven dates). Median size of 

 fish was 176 mm FL (range 128-244 mm) and 98 g 

 eviscerated weight (EW, range 36-289 g). No signifi- 

 cant difference existed between bagged and unbagged 

 fish, and growth was clearly evident over the sam- 

 pling period (Table 1). Few bagged («=3) and un- 

 bagged (n=3) snapper had ruptured swimbladders, 

 and both bagged and unbagged fish had equivalently 

 high indices of stomach eversion ( Table 1 ). Four speci- 

 mens had completely empty stomachs, and all were 

 unbagged fish. Regurgitated prey were typically evi- 

 dent inside the bags of bagged fish and on gill rakers 



Prey volume 



From bagged snapper, we collected an average 1089r 

 more prey (0.75 ±0.57 mL [SD], versus 0.36 ±0.24 

 mL for unbagged fish), on the basis of the unadjusted 

 grand means of all nine date-means weighted by their 

 respective sample sizes (Fig. 1; Table 1). The loga- 

 rithm of prey volume was positively related to the 

 logarithm of fish body weight for bagged fish 

 (P<0.0001) and additionally differed among collec- 

 tion dates (P=0.04; forward stepwise regression, 

 P=0.15 as inclusion level; Proc REG: SAS, 1988; 

 7? 2 =0.98). Date effects thus represented more than 

 just temporal differences in fish size. A fish body 

 weight effect on prey volume was undetectable 

 (P>0.95) for unbagged fish, for which only collection 

 date was important {R' A =0.93). A general date effect 

 increased monotonically over the sampling period. 



Prey volumes for both bagged and unbagged snap- 

 per were evaluated on the basis of bootstrapped es- 

 timates for the last seven sampling dates on which 



