is needed for predicting the sizes of adult 

 runs based on the abundance of young salmon 

 at some time in their life cycle after the 

 greater proportion of highly variable mortali- 

 ties has occurred. 



An estimate of the size of returning adult 

 runs based on the numbers of young fry 

 migrating to the sea in the spring should be 

 more reliable. In the winter, high and often 

 quite variable mortality occurs during egg 

 development and fry emergence. By the time 

 the fry are migrating seaward most of the 

 hazards confronting young fish have already 

 been encountered. Predictions of adult abun- 

 dance based on counts of the young at this 

 stage should have a much lower chance for 

 error. 



This thinking was the basis for the salmon 

 fry counting program that the Bureau initiated 

 in the Sound in 1957, The program had the 

 broad objectives of providing information about 

 the numbers of young salmon leaving the 

 streams, so that variations in fresh-water 

 survival could be detected, and of developing 

 a means of predicting the size of adult pink 

 and chum salmon runs. 



Measuring fry abundance in the Prince 

 William Sound area is largely a matter of 

 sampling the 194 salmon streams that flow 

 into the Sound. 



The specific objectives of the 1957 work re- 

 ported here were as follows: (1) Develop tech- 

 niques for trapping salmon fry as they leave 

 fresh water; (2) determine the numbers of fry 

 migrating from areas upstream from traps in 

 selected streams; (3) field test a method of 

 sampling stream intertidal areas; (4) estimate 

 abundance of fry in the intertidal area; and (5) 

 estimate total abundance of pink and chum 

 salmon fry migrating seaward from all streams 

 entering Prince William Sound. 



EXPERIMENTAL, PROCEDURE 



It is obviously impractical to conduct fry 

 counting operations on all of the 194 salmon 

 streams of Prince William Sound. The prob- 

 lem, therefore, was to select a sample of 

 streams that would provide an unbiased esti- 



mate, with established confidence limits, of 

 the total migration of fry from all streams. 



The experimental design selected in the 

 planning stages was that of stratified random 

 sampling, with proportional allocation of the 

 sample size to the various strata. Streams 

 were grouped, or stratified, by size and by 

 timing of the salmon runs, and those in which 

 migrating fry were to be counted were drawn 

 at random from each stratum.* 



Traps were to be installed in each stream 

 selected and fished throughout the period of 

 seaward migration of fry. The number of fry 

 leaving each sample stream was to be esti- 

 mated by measuring the proportion of water- 

 flow strained by the traps and by conducting 

 marking and recovery experiments. The es- 

 timates were to be inflated for the various 

 strata and combined into an estimate of the 

 total number of fry migrating from above the 

 traps for all of Prince William Sound. 



The numbers of fry produced in intertidal 

 areas below the traps were to be estimated 

 separately and by a different technique. Fry 

 were to be excavated from the gravel of sample 

 streams with the use of a quadrat sampling 

 device. The average number of fry per square 

 yard of gravel was then to be multiplied by the 

 total number of yards in which spawning oc- 

 curred. Intertidal estimates were to be com- 

 bined with above-trap estimates for the grand 

 total of fry for all streams of the Sound. 



It was not surprising that under the difficult 

 field conditions encountered, this sampling 

 plan was not fully implemented. A description 

 of the sampling actually accomplished, details 

 of the traps and the methods of fishing them, 

 and the methods of deriving estimates of the 

 numbers of fry follow. 



Method of Stratifying Streams 



Streams were stratified by the size of the 

 spawning run and the timing of the runs of 

 adult pink salmon. Streams in which the adult 

 escapement had averaged more than 5,000 pink 

 salmon during the 10 years from 1946 to 1956 

 were designated "major," and those with less 



AUlK)ugh not selected at random, a stream at Olsen 

 Bay was included, because studies had been done there 

 previously and facilities were available. 



