FISHERY BULLETIN: VOL. 84, NO. 2 



as used here, refers to time since death rather than 

 time since recruitment. 



The procedures followed here differed from 

 previous CDF&G surveys in that more data were 

 taken than were actually needed for the estimate 

 so that departures from model assumptions could 

 be examined. The additional data enabled simula- 

 tion of the sampling procedure to estimate bias and 

 variances, and allowed us to determine the sources 

 of failure of assumptions. We were also able to 

 develop estimates from which some sources of bias 

 had been removed. 



METHODS 



The study was conducted on a chinook salmon 

 spawning area of a small northern California 



stream, Bogus Creek (Fig. 1). The stream was 

 sampled over a 6.5-mi reach from a counting weir 

 upstream to Bogus School road. Sampling was 

 begun on 15 September 1981, at the very beginning 

 of the spawning run, and discontinued on 12 Novem- 

 ber 1981, by which time very little spawning activ- 

 ity was apparent. The stream was sampled weekly 

 during that period; sampling took 2 d during the 

 peak of the run, with one half of the stream being 

 sampled per day. The stream was sampled by two 

 people walking upstream and capturing with a gaff 

 any carcasses seen. Data on each capture were 

 described as follows: 



Place of capture: Edge top, edge bottom, middle 



top, middle bottom, snagged, dry or buried. 

 Size: Small (<65 cm), medium minus (65-69 cm), 



Klamath River 



Figure 1— Study area in north- 

 ern California. 



262 



