count, was rechecked, and no significant dis- 

 crepancies were found. For the early period 

 161 fish should have been tagged, whereas 

 152 were actually tagged; for the middle 

 period 257 should have been tagged, whereas 

 250 were tagged; and for the late period 214 

 should have been tagged, whereas 176 were 

 tagged. Thus, 632 fish should have been tagged, 

 whereas 578 were actually tagged. Since 62 

 tagged fish were observed returning upstream 

 after dropping down through the weir, these 

 were deducted from the theoretical 632 total, 

 leaving 570 as the theoretical number which 

 should have been tagged, compared with 578 

 actually tagged. 



for a population estimate (table 3). In addition, 

 samples totaling 2,000 live fish, 21 of which 

 were tagged, were also used to determine tag 

 ratios. 



The best estimate of the number of fish 

 actually present in the spawning population 

 above the weir, based on spawning ground 

 recoveries of live and dead fish, was 28,818 

 (rounded to 29,000 in future discussions). This 

 estimate was calculated from the ratio of tags 

 found on the spawning grounds (39) to fish 

 sampled (3,335) and the known number of tags 

 in the lake (337)(tables 2 and 3). 



4. Many tags were lost before surveying 

 This was unlikely because scars were easily 

 seen on the highly colored mature sockeye 

 salmon, and only a few scars were observed 

 in samples. 



5. Many fish passed through the weir 

 without being observed. This did happen, but 

 the reason for it could not be explained. * 



39 337 



3,335- n 



n = 28,818 



Independently supporting this estimate is the 

 fact that approximately 12,600 live and dead 

 spawners were observed on all surveys. Since 

 only a fraction of the fish actually present are 

 observed, the total of 29,000 estimated by tag 

 ratios seems reasonable. 



It was necessary then to estimate the num- 

 ber of spawners actually present, since the 

 recorded count of 4,414 is incomplete. The 

 weir tagging and tag sighting data given in 

 tables 2 and 3 formed a basis for the estimate. 



The tag recovery phase of the program was 

 disappointing because bears removed both 

 tagged and untagged fish and because fish 

 passed through the weir undetected, thereby 

 adding untagged uncounted fish which reduced 

 the expected tagged to untagged ratio. The 

 original plan was to sample only dead fish for 

 tags, since determining tagged to untagged 

 ratios from live fish Is not as reliable as from 

 dead fish (Schaefer, 1951). In the Brooks sys- 

 tem, however, ratios determined irom live 

 counts are preferred because bears mutilate 

 dead and dying fish, sometimes making it im- 

 possible to determine if a fish has been tagged. 

 Twenty tags were actually recovered (exclu- 

 sive of Brooks River), but only 18 were usable 



1 A defect In the sampling trap was discovered and 

 corrected in 1958. 



One additional fact should be mentioned. 

 Population estimates based on tag ratios 

 almost always result in estimates that are 

 much too high. Experiments on the Skeena 

 River, which were very similar to those at 

 Brooks Lake in 1957, resulted in estimates 

 almost double the number of sockeye salmon 

 actually present (Brett, 1952). Our estimate 

 at Brooks Lake may also be high. 



SUCCESS OF SPAWNING 



Three hundred and thirty males and 367 

 females from Up-a-tree, One Shot, Hidden, 

 and Headwater Creeks were cut open and 

 their gonads examined. The fish were recorded 

 as "spawned," "partly spawned," or "un- 

 spawned." Spawned was defined as fewer 

 than 25 eggs remaining or 90 percent of the 

 testes evacuated; unspawned, as 90 percent 

 or more of eggs or testes remaining; and 

 partly spawned, as any Intermediate stage. 

 Spawning success was high: only 20 females 

 and 23 males were upspawned or partly 

 spawned. 



