Mortensen et al.: Growth of Oncorhynchus gorbuscha in relation to marine water temperature, secondary production, and sun/ival 321 



[FL| or total length (TL) depending upon the species). 

 These fish have the potential to be either predators 

 on or competitors with pink salmon juveniles. 



Beginning in 1987, offshore areas of Auke Bay 

 (>100 m from shore) were sampled weekly, at night, 

 with a 6.1-m wide x 3.0-m deep surface trawl (Bax 

 et al.M. All tagged salmon and a sample of up to 100 

 untagged juvenile pink salmon from each tow were 

 retained for length and weight measurements and 

 stomach analysis. All other fish captured in each tow 

 were identified, enumerated, and measured. 



Water temperature at 1-m depth was recorded 

 daily in Auke Bay. Spuhn Island was sampled weekly 

 for littoral epibenthic crustaceans in 1987-1989, 

 with an epibenthic pump (1987) (Simenstad et al.-) 

 or an epibenthic sled (1988 and 1989) (Celewycz and 

 Wertheimer. 1996b). Zooplankton data for the upper 

 5-m and upper 40-m water column in Auke Bay were 

 collected concurrently with our study by University 

 of Alaska researchers (Coyle and Paul, 1990). 



Stomachs of a subsample of untagged juvenile pink 

 salmon retained from beach seine and trawl catches 

 were examined to determine feeding habits. Stomach 

 contents were weighed, and prey items were iden- 

 tified, measured, and counted. A mean volume and 

 dry weight were calculated for each prey category, 

 and the index of relative importance (IRI; Pinkas et 

 al., 1971) was then computed by using the formula 



IKI ^iN+V)F, 



where A^ 

 V 



the numerical percentage; 

 the volumetric percentage of each prey 

 item in the diet; and 

 F = the percent frequency of occurrence of 

 the prey item. 



The importance of individual prey groups were com- 

 pared as a percentage of the total IRI for all prey. 



The maximum residence of tagged fish from each 

 weekly release group in Auke Bay was calculated 

 by subtracting the actual release date for a particu- 

 lar lot of tagged fish from the last recovery date of 

 a tagged juvenile from the same lot of tagged fish, 

 yielding the number of days from release to recap- 

 ture. The weighted mean residence time for each 

 tag release group ( D,.,,^ ) was calculated by subtract- 



Ba.x, N. J., E. O. Salo, B. P. Snyder. C. A. Simfiistad. and 

 W. J. Kinnev. 1978. Salnionid outmigi-ation studies in Hood 

 Canal. Final Report FRI-UVV-7819, Phase III, -January to July 

 1977, University of Washington, College of Fisheries, Fisheries 

 Research Institute, Seattl. WA. 



Simenstad, C. A., C. D. Tanner, and R. M. Thorn. 1989. Estu- 

 arine wetland restoration and monitoring protocol. Report 

 FRI-UW-8918, University of Washington, College of Fisheries, 

 Fisheries Research Institute, Seattle, WA. 



ing the weighted mean release date for each weekly 

 release group from the weighted mean recovery date 

 for each release group. The weighted mean release 

 date (D,.j,/ ) for weekly release group i was calculated 

 from the formula 



D. 



rel, 



I 



(/ = ! 



E<"^<i 



(/ = n 



I^<, 



d=l 



where ii - the number of release days per week; 

 d^^i = the actual release day (Julian date); and 

 Nj^^ = the number of tagged fish from that 

 day that were released. The weighted 

 mean residence for each weekly group 

 was then calculated by subtracting the 

 weighted mean release date of a juvenile 

 for a given week from the weighted mean 

 recovery date. The weighted mean recov- 

 ery date ( D,^,^. ), for a particular release 

 group was calculated as 



IL'^rec^^: 



D .. =^ 



d=l 



I^^. 



where « = the number of recovery days per week; 

 d ,.^.^ = the actual recovery day (Julian date) of a 



tagged juvenile; and 

 ^d,,, = the number of tagged fish recovered on 

 day d. 



The weighted mean residence time ( D,.^,^ ) was regressed 

 against release date for each year Analysis of covari- 

 ance ( ANCOVA) was used to detennine whether signif- 

 icant differences existed among years, followed by the 

 Newman-Keids multiple comparison test to determine 

 if significant differences (a=0.05) existed within years. 

 The instantaneous growth rate, g. for individual 

 tagged pink salmon juvenile was calculated as 



W.-, = W,exp«', 



where / = the period in days from the mean 

 release date, over which the growth 

 rate was calculated; and 



Wj and W., = the fish weight at the beginning and 

 end of the period, respectively. 



In addition the natural log of the weight of tagged 

 fish at capture was regressed against days from 



