DeMartini et al : Pressure effects on regurgitation of Pristipomoides filamentosus 



251 



As part of our studies of the foraging and distribu- 

 tion of juvenile pink snapper, we have been develop- 

 ing methods to quantify stomach contents with ac- 

 ceptable accuracy and precision. In this paper we de- 

 scribe and evaluate a method of collecting juvenile 

 snapper for diet analysis that minimizes the prob- 

 lem of subsurface regurgitation of food as specimens 

 are retrieved from depth. 



Methods 



Fish collection 



Juvenile snapper were caught by hook and line from 

 near-bottom depths of 60-90 m, about 3-4 km off- 

 shore of Kaneohe Bay, windward Oahu, Hawaii, on 

 nine dates during the period February-August 1994. 

 All fish were captured between about 0900 and 1500 

 h. On each date, at 15-18 m below the sea surface, 

 about one-half of all fish (randomly selected) were 

 "bagged" by scuba divers (using NOAA NITROX II 

 [64% N„, 36% 2 ] to extend diving periods approxi- 

 mately twofold at these depths). Divers observed 

 fish — underwater visibilities consistently exceeded 

 30 m — as they were reeled up to 15-18 m and subse- 

 quently to the sea surface and noted whether speci- 

 mens regurgitated food just before or during the bag- 

 ging process; if they did regurgitate, these fish were 

 left "unbagged" (see below). Median time for collec- 

 tion (1055 h) was the same for both bagged and 

 unbagged fish. Bagged fish were sealed live in indi- 

 vidually marked canvas bags before they were re- 

 trieved the remaining distance to the surface. The 

 other half of the fish (unbagged) were brought di- 

 rectly to the surface by the fishing line. All fish were 

 reeled in at a rate of about 30 m per minute (Haight 

 et al., 1993b I. Once aboard ship, unbagged fish were 

 also packaged individually in marked canvas bags. 

 Fish were stored on ice in a cooler until they were 

 frozen intact in their marked bags at the NMFS Ho- 

 nolulu Laboratory, 2-8 h after capture. 



Stomach content analyses 



Each frozen fish was thawed, removed from its bag, 

 and the bag everted in the laboratory. The external 

 body surface of each specimen and the inside sur- 

 face of the bag were rinsed into a black paraffin-bot- 

 tom dissecting tray. Any prey fragments in the fluid 

 were collected. Gill rakers, esophagus, and pharynx 

 (including vomerine and jaw teeth) of the specimen 

 were examined for prey. Ruptured swim bladders 

 were noted, and extent of stomach eversion accord- 

 ing to a stomach eversion index (EvI) was recorded 



(on a scale from 1 [not everted] to 2 [partially everted] 

 to 3 [fully everted]). Fish were then measured (mm 

 fork length [FL]). Lastly, the stomach was excised 

 and its contents saved, along with any prey items 

 that were free in the coelom as the result of stomach 

 rupture. All prey (regurgitated and present in the 

 stomach) were pooled. Prey were then blotted damp 

 on a paper towel, and a displacement volume (0.01 

 mL) was measured. Stomachs and prey were fixed 

 in 3.7%. formaldehyde (sea water buffered) for 1-2 

 months, then preserved in 60% EtOH. The eviscer- 

 ated wet weight (EW) of each snapper was recorded 

 to the nearest gram. 



Prey were examined under a dissecting microscope 

 at 8-50xand classified (when possible) to the level 

 of suborder or order. The length ( longest axis) of each 

 prey item was recorded to the nearest millimeter. 



Identifiable prey were divided a priori into four 

 major types (Bowen, 1983) on the basis of probable 

 microhabitat: 1) crustaceans — mostly shrimps and 

 stomatopods of indeterminate epibenthic or lower 

 water column habitats; 2) benthos — primarily dem- 

 ersal octopods, echinoids, and microgastropods rep- 

 resenting mobile epifauna; 3 ) nekton — actively swim- 

 ming water column fishes and squids; and 4) jellies — 

 weakly mobile, gelatinous salps and heteropods. 



Statistics 



A nonparametric bootstrapping procedure was used 

 to evaluate mean prey volume in bagged versus 

 unbagged fish because we were interested in quan- 

 tifying the magnitude of possible differences in prey 

 volume between the two treatments, even though 

 sampling dates were too few to evaluate normality 

 accurately. The variables used for bootstrapping 

 (n = l,000 iterations; Manly, 1991) were 1) the differ- 

 ences between the date-specific mean prey volumes 

 for bagged and unbagged fish. Because sampling 

 design alone might have incompletely controlled for 

 body size and date effects on prey volume, we fur- 

 ther evaluated 2) the differences between predicted 

 and observed prey volumes in unbagged snapper. 

 Predicted values for unbagged fish were calculated 

 on the basis of prey volumes measured for bagged 

 fish of the same weight and date, because prey 

 volumes were positively related to body weight for 

 bagged snapper only ( see Results section ). Predicted 

 values for bagged fish were derived from stepwise 

 multiple regression of prey volume on sampling 

 date and body weight. For both 1) and 2), date-means 

 were weighted by the number of sample fish on that 

 date. 



Standard nonparametric tests (Siegel and Cast- 

 ellan, 1988; SAS, 1988: Proc NPAR1 WAY, Proc FREQ ) 



