360 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



the migrants quickly and efficiently, (3) audible 

 timers to limit the counting periods precisely and 

 eliminate the need of watching a clock, (4) guid- 

 ing devices to lead the migrants over the panels, 

 and (5) spotlights to assist in night counts. 



ESTIMATION OF THE RUNS 



The sampling procedure was designed to ob- 

 tain a reasonably accurate estimate of the total 

 escapement while staying within the limits of both 

 manpower and budget. Most of the information 

 on which the initial sampling plan was formulated 

 was known from previous observations of red sal- 

 mon moving upriver and through weirs. Usually 

 the major peak of the migration occurred in a day 

 or two, although it was often preceded or fol- 

 lowed by minor peaks. There were usually 

 marked variations in numbers of fish from day 

 to day and throughout a given day. In some lo- 

 cations near the fishery there were also marked 

 changes in the numbers with the opening and 

 closing of the fishing periods or with changing 

 tides. It has since been noted at Igiugig on the 

 Kvichak Eiver that the heaviest runs usually ap- 

 pear along one bank at a time, with comparatively 

 small numbers of fish passing along the opposite 

 shore. The major path of the run changes er- 

 ratically during the season and from year to 

 year, but once a heavy run develops along one 

 bank, it is usually maintained for several days. 



A typical migration up the Kvichak River was 

 counted continuously on July 16, 1955, in front 

 of tower No. 1 (fig. 6). The fluctuations during 



^ 1600 



(J I400' 



I 1200 



5 



^ 1000 



soo 



ft: 600 



Meon of 507 

 tish/lOi 



0000 0200 0400 0600 



1000 i:oo 1400 leoo isoo 2000 2200 



Hours 



Figure 6. — Numerical fluctuations in migrating red sal- 

 mon, from lO-minute continuou.s counts taken on July 

 IG, 19.")5, at tower No. 1. 



daylight ranged from about two-thirds of the 

 mean of 507 fish each 10 minutes to about one 

 and one-half times the mean. With such varia- 

 tions, it is desirable to sample every day and 

 within each day to use a large number of short 

 counting periods equally spaced in a .systematic 

 fashion. 



Since the magnitude of the escapement fluctu- 

 ates greatly, even in successive 10-minute counts, 

 the estimates calculated from the samples are 

 basically point estimates. A single count depicts 

 accurately only the number of fish that happens 

 to be passing at the time. Frequent counts depict 

 accurately only the number of fish passing when 

 the counts were taken, yet they are closely related 

 to the total numbers in the fluctuating population. 

 As a result, the daily estimates obtained from the 

 counts approximate the numbers of migrating sal- 

 mon, with the accuracy depending on the magni- 

 tude and frequency of the fluctuations and the 

 frequency and duration of the sample counts. 



Method of Sampling 



The basic sample unit used in 1957, 1958, and 

 1959 consisted of 10-minute counts taken system- 

 atically each hour from each tower. The initial 

 sample counts in 1955 were of 2 hours' duration and 

 taken every 4 hours. In 1956, counts were even- 

 tually reduced to 40 minutes and were taken every 

 3 hours. But the 10-minute counts were found 

 most practical to obtain and to use in calculating 

 the estimate. Any sampling procedure will be 

 improved if it is flexible enough to permit in- 

 creases or decreases in the degree of coverage with 

 the intensity of the migration. Consequently, at 

 the beginning and end of the migration the counts 

 were taken every 2, 3, or 4 hours. Also, at the 

 peak of the migration or if a heavy run com- 

 menced along one bank, counts were increased 

 to 15 or 20 minutes. This flexibility placed em- 

 phasis on increasing the accuracy of the total esti- 

 mates by concentrating the counts on the periods 

 with the heaviest escapements. 



Counts were taken by natural light as early in 

 the morning and as late at night as possible. The 

 hours of darkness normally extend from 2300 to 

 0200, but often vary from day to day depending 

 on the extent of sky overcast and the season. Ac- 

 curate night counts depend on seeing the fish by 

 means of artificial light from spotlamps. The 

 intensity of these lamps was controlled by a rheo- 



