made to estimate the original number. This 

 procedure prevented errors of duplication 

 but made the totals minimal. 



Because zooplankters usually were too nu- 

 merous to count, their numbers were esti- 

 mated as follows: The bottom of a petri dish 

 was marked into 1-cm. squares with a dia- 

 mond-tipped pencil; plankters were uniformly 

 distributed over the dish; and at least four 

 squares were counted through a binocular 

 microscope. The average of the counted 

 squares was then multiplied by the area of 

 the petri dish (64 cm. 2) to estimate the total 

 numbers. 



To compare changes in diet, the season 

 was arbitrarily separated into three periods 

 (July, August, and September), and food of all 

 gill net-caught fish from each period was 

 combined (fig. 24). The following significant 

 facts are apparent: 



1. Contrary to the usual belief that sock- 

 eye salmon are primarily plankton feeders, 

 insects were found in an even greater per- 

 centage of stomachs examined during the 

 season than was plankton — 30.2 as opposed to 

 26.2. 



2. There was a well-defined progressive 

 seasonal shift in the diet from insects during 

 the first part of the summer to plankton during 

 the latter part of the summer. This same 

 shift in diet was also found by Russian inves- 

 tigators at Lake Dalnce, Kamchatka (Krogius 

 and Krokhin. 1956a). 



indicating that availability of insects during 

 a given year may be more important to the 

 juveniles' well-being than availability of plank- 

 ton. 



2. Winged insects, which make up the bulk 

 of the diet, are generally more abundant over 

 the littoral areas adjacent to land. Thus, 

 preference (or availability) of insects near 

 shore during the early summer is conducive 

 to concentrating sockeye salmon near the 

 shoreline (table 10). 



3. The simultaneous movement to mid- 

 lake areas, decrease in insect feeding, in- 

 crease in plankton feeding, and increased 

 incidence of empty stomachs all contribute 

 to slower growth and early formation of an 

 annulus. 



4. Availability is probably the governing 

 factor in the lower incidence of insect feeding 

 as the season progresses. Gross obsers'ations 

 indicate that the peak of insect production is 

 in June, coinciding with the period of maximum 

 daylight. By October very few winged insects 

 are evident at Brooks Lake. 



5. Sockeye salmon select individual food 

 organisms, even species of plankton. In many 

 instances stomachs of fish caught at the same 

 time and place contained entirely different 

 organisms. Sometimes a stomach from a 

 fish would contain only one species of plank- 

 ton in large numbers, while a fish captured 

 in adjacent meshes of a gill net would contain 

 another species. 



3. Incidence of empty stomachs increased 

 progressively through the season. 



4. A few sockeye salmon were piscivorous 

 and even cannibalistic, as indicated in the 

 "other" category in figure 24. which com- 

 prised mostly fish remains. 



Speculation on the implication of these facts 

 leads to some interesting tentative conclu- 

 sions; 



1. The period of insect feeding corre- 

 sponds to the period of most rapid growtii of 

 juveniles during the spring and early summer. 



The shift from insects to plankton through 

 the season in terms of total numbers is shown 

 in table 12. Total numbers of organisms are 

 shown rather than frequency of occurrence as 

 in figure 24, The overall ratio of individual 

 Insects to plankton organisms was 1:25. The 

 percentage fullness of stomachs through the 

 season was also measured (fig. 25). Sockeye 

 salmon ate less as the season progressed, with 

 the Incidence of to 10 percent fullness more 

 than doubling, from 22 percent In June and 

 July to 59 percent In September. This slacken- 

 ing of feeding further explains the late summer 

 decrease In growth rate and fall annulus 

 forniatlon. 



36 



