608 



Fishery Bulletin 104(4) 



Figure 2 and the number of samples collected 

 from all sources during all seasons is shown 

 in Figure 3. 



In addition to the stomach contents col- 

 lected and analyzed for the present study, 

 diet data collected regionally were drawn 

 from the NMFS-NEFSC bottom trawl survey 

 database. A full description of the survey 

 design and collection methods used by the 

 NEFSC can be found in Azarovitz et al.'' 



Diet analysis 



Winter 



Spring Summer 



Season 



The majority of samples obtained from fish- 

 ery-independent sources were processed 

 onboard research vessels. However, only cur- 

 sory assessments of prey type and mass could 

 be made at sea; therefore samples were frozen 

 and transported to the laboratory for a more 

 thorough examination. Each fish was weighed 

 and measured (fork length for bluefish, total 

 length for goosefish, silver hake, and summer 

 fiounder), stomachs were removed, prey items 

 were weighed to the nearest 0.01 gram and identified 

 to the level of species whenever possible. With the aid 

 of a dissecting microscope, otoliths, beaks, and other 

 hard parts were recovered in order to classify species. 

 When stomach contents could not be identified because 

 of advanced stages of digestion, they were recorded as 

 "unidentified animal remains." As with all diet stud- 

 ies, there is always a chance that a portion of samples 

 are biased because fish may feed while in the net. This 

 problem was addressed by excluding fish where prey was 

 found (undigested) in the mouth and esophagus or where 



Figure 3 



Frequencies (in days) of samples of bluefish iPoniatomus !taltat?'ix), 

 goosefish iLophius americanus), silver hake [Merluccius bilinearis), 

 and summer flounder iParalichthys dentatus) collected from fish- 

 ery-dependent and fishery-independent sources. No samples were 

 collected for bluefish iPomatomus saltatnx) during the winter and 

 spring seasons. 



Azarovitz, T., S. Clark, L. Despres. and C. Byrne. 1997. The 

 Northeast Fisheries Science Center bottom trawl survey 

 program. 22 p. ICES Council Meeting 1997/Y:3.3. 



B Bluefish BGoosetish nSilverhake D Summer flounder 



1.0 1 



Winter 



Spnng 



Summer 



Fall 



Season 



Figure 2 



The proportion of samples of bluefish iPomatomus saltatrix). 

 goosefish {Lophius americanus), silver hake [Merluccius bilin- 

 earis), and summer flounder iParalichthys dentatus) specimens 

 collected from fishery-independent sources. No samples were 

 collected for bluefish during the winter and spring seasons. 



there was a clear difference in digestion stage among 

 prey items found in the stomachs. 



Three size classes, covering the total range of lengths 

 offish collected, were chosen for each species as follows: 

 bluefish — small (200 to 450 mm fork length), medium 

 (451 to 550 mm), and large (>551 mm); goosefish — small 

 (50 to 250 mm total length), medium (251 to 450 mm), 

 and large (>451 mm); silver hake — small (50 to 199 

 mm total length), medium (200 to 299 mm), and large 

 (>300 mm); and summer flounder — small (250 to 399 

 mm total length), medium (400 to 549 mm), and large 

 (>550 mm). Seasonal time periods were defined as win- 

 ter (December-February), spring (March-May), summer 

 (June-August), and fall (September-November). 



Statistical analyses 



The Kruskal-Wallis test was used to test seasonal 

 and size-based differences in predation on squid. 

 A nonparametric test was chosen because of its 

 robustness to assumptions of normality and skew- 

 ness (Quinn and Keough, 2003). Preliminary anal- 

 yses indicated that none of the data sets fitted the 

 normal distribution and, owing to large numbers 

 of zero values (representing the absence of squid 

 in the diet), could not be transformed. Two sets of 

 tests were performed on the data: first an "aggres- 

 sive" set of tests were run by using individual fish 

 as the sampling unit to give the maximum number 

 of degrees of freedom; second, a "conservative" 

 set of tests were run by using either stations (for 

 data collected from fishery-independent sources) or 

 cruises (for data collected from commercial fishery 

 sources) as the sampling unit. The percent mass 

 of squid in the diet of each fish (aggressive test) 

 or pooled fish per station or cruise (conservative 

 test) was determined and expressed as a proper- 



