FISHERY BULLETIN: VOL. 87. NO. 4, 1989 



where crab predation occurred, and whether this 

 changed with time in response to environmental 

 variables; 2) changes in the percentage by 

 weight and size distribution of crab in the diet by 

 year; and 3) total amounts of crab consumed by 

 the Pacific cod population for each year calcu- 

 lated from food habits data, daily ration, and 

 Pacific cod population abundance estimates. 

 Estimated removals of crab by Pacific cod for 

 each year were compared with the size distribu- 

 tion and population levels of crab estimated from 

 annual research surveys to determine whether 

 cod exerted density-dependent control over crab 

 population size. 



METHODS 



Sample Collection and Laboratory 

 Analysis 



Stomachs were collected from 4,023 Pacific 

 cod (30-107 cm fork length (FD) during the 

 three years (1981, 1984, and 1985) in the eastern 

 Bering Sea (Fig. 1, Table 1). Samples were 

 taken from May through September using bot- 

 tom trawl gear on research and commercial fish- 

 ing vessels. Sampling occurred throughout the 

 24 h day in 1984 and 1985 and from 0600 to 2000 

 Alaska daylight time in 1981. Stomachs were 

 removed at sea and placed in cloth bags labelled 

 with information regarding the location of cap- 

 ture and the length, sex, and sexual maturity of 

 the fish. Individual fish weights were calculated 

 using a length-weight relationship developed for 

 Pacific cod in the eastern Bering Sea (Bakkala et 

 al. 1986). Fish showing evidence of regurgitation 

 (i.e., food in the mouth or throat, or a flaccid 

 stomach) were not included in the sample. 

 Stomachs were preserved in 10% formalin and 



later transferred to 70% ethyl alcohol. Contents 

 were identified to the lowest taxonomic level 

 possible and enumerated. Wet weights were re- 

 corded after the contents were blotted with 

 paper towels. If carapaces were intact, snow 

 crabs in the stomachs were measured to the 

 nearest millimeter carapace width (CW), and 

 king crabs were measured to the nearest milli- 

 meter carapace length (CD. 



Data Analysis 



Pacific cod were divided into two size groups 

 for data analysis: 30-59 cm FL and >60 cm FL. 

 Previous studies (Livingston et al. 1986) show 

 that cod become increasingly piscivorous beyond 

 60 cm FL, and mean stomach content weight as a 

 percentage of body weight is also much larger 

 for cod >60 cm in length. Thus, the food habits 

 and daily ration need to be examined separately 

 for the two size groups of cod. 



Logistic regression, using the BMDPLR 

 routine in the BMDP statistical software pack- 

 age (Dixon 1983), was performed to determine 

 which major factors (predator size and year) 

 were important in describing variation in preda- 

 tion by cod on a particular crab species. The 

 dependent variable was the presence or absence 

 of a crab species in cod stomachs from two cod 

 size groups (Size 1 = 30-59 cm. Size 2 = >60 cm) 

 during the three sampling years (1981, 1984, and 

 1985). The most parsimonious model was chosen 

 for describing predation on each crab species 

 using the criteria that 1) only factors that re- 

 sulted in a significant improvement (P < 0.05) in 

 model fit be added and, 2) the overall model 

 goodness of fit chi-square be nonsignificant (P > 

 0.05), indicating that the model provided a good 

 fit to the data. In addition, BMDPLR permitted 



Table 1 . — Stomach sample collection information of Pacific cod taken in 1 981 . 1 984, and 1 985 in 



the eastern Bering Sea. 



'ADT is Alaska daylight time. 



808 



