364 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



in the water which then flows downriver and re- 

 duces visibility at the tower sites. Occasionally 

 the point is reached where counts are not possible. 

 This is discussed as high turbidity, under the sec- 

 tion on "Accuracy of the Estimates.'" 



"WHien the water is moderately turbid, the vari- 

 ation between comparative counts might be ex- 

 pected to increase. Yet the data presented in 

 table 4 show that the variation between compara- 

 tive counts taken in moderately turbid water was 

 less (-1-1.7 percent) than those obtained in clear 

 water (-3.0 percent). This suggests that mod- 

 erately turbid water did not increase the counting 

 variation between two counters. It does not indi- 

 cate that both counts were as accurate in turbid 

 water as in clear water, especially on the fringe 

 areas of the migration paths. 



Effect of Disrupted Migration Pattern 



Migration past a tower is momentarily dis- 

 rupted as the fish scatter into deep water when 

 the skiff bearing the counter arrives. While the 

 fish soon re-form their ranks and resume an ap- 

 parently normal migration pattern, it is possible 

 that commencing the counts as soon as the counter 

 is ready would inject directional errors. 



A test of the timing of the coimts to the time 

 of the counter s arrival was made in 1957 by com- 

 paring the first and second 5-minute counts of 

 841 samples (table 5). In most instances these 

 counts were started as soon as the counter had 

 tied the skiff, climbed the tower, and readied his 

 equipment, a period of 1 or 2 minutes. Since the 

 first count was higher than the second almost one- 

 half the time, i.e., neither consistently higher nor 

 lower, the counts were probably not biased. The 

 time required in preparation to commence the 

 counts, therefore, is evidently sufficient for the 

 migration to assume a normal pattern. 



Table 4. — Effect of turbid water on count variation 



ACCURACY OF ESTIMATES 



Estimates of the total migration are subject to 

 statistical error because of periods when the conti- 

 nuity of the sample counts is interrupted, and 

 because of fluctuations in abundance of fish from 

 one counting period to another. At present we 

 can only guess at possible bias that may accrue 

 because of interruptions in the counts, but we can 

 estimate precisely the possible statistical error for 

 fluctuations of abundance. For example, it is pos- 

 sible to calculate confidence limits for each annual 

 escapement to determine the reliability of the 

 sampling program. In addition, various mathe- 

 matical tests can be applied to determine the effects 

 of changes in the length of the samples and the 

 interspacing periods. 



Effect of Interruptions in the Counts 



Interruptions in counting have occurred because 

 of high turbidity when no fish can be seen, also 

 because of darkness, especially before satisfactory 

 lighting was developed. 



Highly turbid water occurred 10.9, 3.7, 3.1, 6.2, 

 and 5.5 percent of the time in the years from 1955 

 to 1959, respectively. Once the river becomes 

 turbid it normally remains in that condition for 

 at least 24 hours, because the water takes about 

 8 hours to clear after the east wind ceases. The 

 periods when high turbidity prevailed, in relation 

 to the magnitude of the escapement each year, are 

 shown in figure 9. Only in 1957 did the water 

 turn highly turbid when a heavy migration of red 

 salmon was passing the towers. 



Beach seine hauls have been obtained occasion- 

 ally to provide relative indications of migration 

 intensity in turbid waters. These hauls are prob- 

 ably more effective in turbid than in clear water 

 because of inability of the fish to see the net. Yet, 

 catches were low, indicating a low migration in- 

 tensity. A systematic comparison of liauls in clear 

 and in turbid water was made in 1958, where the 

 average catch per haul along the right bank was 

 compai-ed with the estimated number of fish 

 migrating past tower Xo. 1 (fig. 10). The curve 

 formed by this relationship indicated that few 

 fish were migrating in the higlily turbid water. 

 Tlierefore, calculations of the migration for the 

 period of turbid water, based on an average of low 

 counts immediately preceding and following the 

 turbid period, weie basically correct. 



