206 



Fishery Bulletin 104(2) 



the chorion was relatively intact. The presence of empty 

 chorions was noted but these were not included in the 

 analysis because of their potentially longer residence 

 time in the stomach of predators (Hunter and Kimbrell, 

 1980). As it turned out, empty chorions occurred in less 

 than 2'7f of capelin with fish eggs present in their gut. 

 For every fifth capelin specimen, all prey items were 

 identified according to broad taxonomic groups: Co- 

 pepoda, Amphipoda. Euphausia, Pteropoda, molluscan 

 veligers, brachyuran zoeae, Larvaecea. Cladocera, Chae- 

 tognatha, Isopoda, and fish eggs and larvae. The num- 

 ber of individuals was determined for each taxonomic 

 group from the entire stomach content and the total 

 weight for each group was determined to the nearest /.ig 

 by using a Cahn microbalance. This protocol was fol- 

 lowed to provide a complete assessment of the presence 

 of fish eggs in the gut and to provide a representative 

 assessment of the patterns of prey composition from 

 each site to confirm that general feeding patterns were 

 consistent with results from previous studies. O'Driscoll 

 et al. (2001) had performed a detailed analysis of feed- 

 ing patterns of capelin in Newfoundland waters and I 

 perceived no need or advantage in repeating the work. 

 The complete analysis of stomach contents was per- 

 formed on 218 capelin specimens. 



All fish eggs and larvae were sorted from plankton 

 samples and identified to the lowest taxonomic level 

 possible. Because of difficulties in differentiating early 

 developmental stages of two species groups, eggs were 

 lumped into either of two groups: 1) CHW (namely, cod 

 (Gadus morhiia L. ), haddock iMelanogrammus aeglefi- 

 nus L.) and witch flounder (Glyptocephalus cynoglossus 

 L.) or 2) CYT (cunner (Taiitogolabrus adspersus L.), 

 yellowtail flounder [Limanda ferriiginea, Storer 1839)). 

 Eggs were most likely those of cod, for the CHW cat- 

 egory, and cunner, for the CYT category because only 

 larvae of those species were found in other elements of 

 the research program. A subsample of other zooplank- 

 ton, yielding 200-300 specimens, was identified to the 

 lowest taxonimic level possible. 



For each capelin, the total gut fullness index (GFI) 

 was estimated as GFI=WJW, where W,=the weight of 

 stomach contents (g); and W=is the weight of the fish 

 (g), based on the local bias-corrected length-weight rela- 

 tionship (log,|j W=-6.44+3.38 logjo TL, where W=weight 

 in g and T'=total length in mm; r^ = 0.98; and the stan- 

 dard error of the intercept and slope are 0.049 and 

 0.023; Mowbray2). 



Analysis 



was determined. These data, arcsine-square-root trans- 

 formed, were contrasted using principal component anal- 

 ysis in which the proportion of prey type, log-transformed 

 capelin length and GFI were included to determine if the 

 observations from this study were consistent with those 

 from previous ones. 



Frequency of fish eggs 



The probability distribution of egg numbers per gut 

 was contrasted with the expectations based on a Pois- 

 son process. Encounters between predator and prey can 

 be described by using a modification of Gerritsen and 

 Strickler's (1977) equations where 



■^■R~ -^pr,; VU~ + V'^, 



(1) 



where 7? = the reactive distance (m); 



A .,,^, = the density of prey (/m'^); and 

 u and V = the swimming speeds of predator and prey 

 (m/h), which yields an encounter rate A (per h). 



Huse and Toresen (2000) used laboratory estimates 

 of the various parameters to estimate encounter rates 

 between juvenile herring iClupea harengus L.) and cap- 

 elin larvae. Instead of using their approach, I chose to 

 reduce Equation 1 to its simplest form, such that 



X = K A„ 



(2) 



where K = the effective volume swept (m^/h) during 

 the period during which eggs will remain 

 discernible in the gut of capelin; and 

 A . - the observed density of eggs (/m^) in the 

 water column from plankton samples. 



The resulting probability of finding A'^ prey in the gut of 

 capelin then becomes 



P(Ar) = -.e-\ 



where A = the mean encounter rate. 



(3) 



Stomach content composition 



Mean number of eggs per stomach (A''), taken as a mea- 

 sure of encounter rates for each four-hour sampling 

 interval, was estimated by fitting a generalized linear 

 model with Poisson error structure using a log link 

 function by maximum likelihood (GENMOD procedure, 

 vers. 8, SAS Inc., Gary, NC) to the data from all 4-hour 

 sampling intervals (T) from all locations, with TL of 

 individual capelin as a linear covariate. 



For each capelin on which the full stomach contents 

 were analyzed, the number of each major prey taxon 



2 Mowbray, F. 2005. Personal commun. Fisheries and 

 Oceans, St. John's, Newfoundland, Canada. 



f(N)^h^,+h{r + h.{rL. 



(4) 



The value oi N reported in the "Results" section is the 

 least squares adjusted mean, which represents the value 

 of the average length of capelin from field collections. 

 We attempted to include site into the analysis but the 



