RED SALMON COUNTS FROM OBSERVATION TOWERS 

 Table 5. — Relation of 1st 5-minute counts to 2d 3-minute counts, Kvichak River, 1957 



365 



Night counts were not taken in 1958 because of 

 difficulty in counting the fisli and the questionable 

 accuracy of counts made under floodlights. In- 

 stead, estimates were made on the assumption 

 that the migration was constant between the last 

 evening count and the first morning count. The 

 total time lost to sampling by omitting night 

 counts each da}- was 4 hours, 2300 — 0200. Esti- 

 mates of the escapements based on systematic 

 night counts in 1957 and 1959 from tower No. 1 

 were compared with those calculated by averaging 

 the 2200 and 0300 samples (table 6). In both 

 instances the escapements test -calculated by omit- 

 ting night counts were high, 6.8 percent for the 



Highly turbid 

 woter 



^ 



St 



FiGVRE 0. — Relationship of periods of high turbidity to 

 red salmon escapements, Kviohak River, 1!).").5 through 

 1959. 





70-1 



60- 



50 



S 40- 



^•fi s 



^ g 30- 



20 



10 



• Houls made rn clear water 



o Houls mode in highly turbid waters 



ie_ 



—\ — I — 

 25 



50 



75 



Average Number of Red Salmon/ Seine Haul 



Figure 10. — Relation between average numbers of red 

 salmon caught per seine haul and the calculated number 

 of fish pa.ssing tower No. 1. Data computed on a ',i-day 

 basis, Kvichak River, Jul.v 7-17, 1958. 



larger 1957 run and 34.6 percent for the smaller 

 1959 run. Consequently, it is necessary to obtain 

 night counts whenever possible to maintain a high 

 degree of accuracy in the estimates. 



Population Fluctuations 



Fluctuations in numbers of migrating fish from 

 season to season, from day to day, and from hour 

 to hour, are characteristic of Kvichak River 

 salmon runs. Seasonally, the fluctuations have 

 ranged from 14 to 91/2 million fish. Dailj' fluctu- 

 ations were clearly defined in small escapements, 

 less defined in large runs. 



However, fluctuations taking place between 

 counts are of the greatest importance to the esti- 

 mates of the escapement. The greater the fre- 

 quency and magnitude of fluctuations from one 

 sample to the next, the more the calculated esti- 

 mates are likely to deviate from the true popula- 

 tion passing upriver. Fluctuations from sample 

 to sample and between samples are the result, to 



