This difference may be caused by the relative 

 sizes of the rearing areas of the two systems, 

 because the slopes of the regressions of smelt 

 size on numbers of parent spawners per square 

 kilometer of nursery area are similar for the 

 two systems. 



A regression of ocean survival on smolt size 

 with a positive slope of about 0.25 can be cal- 

 culated from data compiled by Ricker (1962). 

 Thus, although it appears that in the Kvichak 

 system an escapement of about 10 million spawn- 

 ers will produce the greatest number of smolts 

 (age groups I and II combined),'" the survival 

 of these smolts after they leave the lake will be 

 reduced because the average size of fish in each 

 age group is less than the size of the fewer smolts 

 produced by a smaller escapement. Clearly, the 

 optimum escapement for peak years in the Kvi- 

 chak system must be determined by maximizing 

 the return as it is influenced by numbers and 

 quality of smolts produced. The optimum escape- 

 ment mu.st be less, therefore, than that which 

 produces the largest number of smolts. 



Another factor also indicates a need for a 

 lower escapement. The regression line in figure 

 17 has one aberrant point, namely, the length of 

 age I smolts from the 1957 escapement (2,843,- 

 000) which followed the large escapement of 



1956 (9,443,000). Although the escapement in 



1957 was comparatively small, the mean length 

 of age I smolts from the 1957 escapement was 

 less than that of age I smolts from the large 1956 

 escapement. This discrepancy is interpreted to 

 mean that as young-of-the-year, the age I smolts 

 of the 1957 e.scapement competed for food with 

 the abundant yearling fish from the 1956 escape- 

 ment which migrated as age II smolts. Optimum 

 escapement cannot be determined, therefore, by 

 considering one year alone — possible effects of 

 competition between progeny of successive es- 

 capements must be carefully considered. 



Naknek and Ugashik Systems 



Little indication exists for the Naknek and 

 Ugashik systems that the average size of age I 

 smolts is correlated with the number of adults in 

 the parent escapement (figs. 18 and 19) . The age 

 II smolts show a similar lack of correlation (data 

 not shown) . Scales from smolts indicate that the 

 variations in size of age I fish at the time of sea- 



110 



£ 105 



100 - 



< 



o 



I 

 I- 

 o 

 ■z 

 u 



95- 



90 



0.5 10 1.5 2.0 2.5 



FISH IN PARENT ESCAPEMENT (MILLIONS) 



Figure 18. — Relation between mean lengths of age I 

 sockeye salmon smolts and number of adults in parent 

 escapements, Naknek system, 1956-61. The number be- 

 side each point indicates the parent escapement year. 



' This estimate is based un data collected through 1963 (see footnote 



16). 



0.5 1.0 1.5 2.0 2.5 



FISH IN PARENT ESCAPEMENT (MILLIONS) 



Figure 19. — Relation between mean lengths of age I 

 sockeye salmon smolts and number of adults in par- 

 ent escapements, Ugashik system, 1956-61. The num- 

 ber beside each point indicates the parent escapement 

 year. 



ward migration are caused to a great extent by 

 variations in growing conditions in the lakes just 

 before the smolts leave. 



Mortality of sockeye salmon in the ocean is 



SOCKEYE SALMON IN MAJOR RIVER SYSTEMS IN SOUTHWESTERN ALASKA 



445 



