Terminations of the migrations were not ob- 

 served in any of the streams except Cannery 

 Creek and Olsen Bay, where they ended on 

 June 17 and July 4, respectively. After these 

 dates, only insignificant numbers of fry were 

 taken. 



Proportion of Fry Captured 

 by Traps 



Comparison of results of marking experi- 

 ments with volume studies suggests that where 

 water velocities were high, there was better 

 agreement between the two methods of deter- 

 mining the proportion of fry captured by traps 

 than when velocities were low (table 2). The 

 differences between percent of water volume 

 strained and percent of fry recaptured were 

 large in some instances, and a choice of method 

 and the resultant data had to be made. One 

 explanation of the discrepancies between the 

 two methods derives from inadequate sampling. 

 At some of the study streamsonly one marking 

 and recovery experiment was conducted, which 

 was probably insufficient to establish the re- 

 capture rate accurately. 



Behavior of the fry in relation to the traps 

 offers an explanation of why the percent of total 

 streamflow strained by the traps was higher 

 than the percent of marked fry recovered. Both 

 pink and chum salmon fry avoided trap instal- 

 lations when stream velocities were less than 

 2 feet per second. Frequently fish were ob- 

 served suddenly increasing their swimming 

 speed when they neared the traps. On a few 



occasions they would swim downstream and 

 then suddenly turn 90 degrees when they 

 neared the traps. These reactions may have 

 been caused by the complete trap installation 

 instead of an individual trap, but no fish show- 

 ing this avoidance reaction was observed to 

 enter a trap. It is assumed that marked fry 

 behaved the same as unmarked fry; therefore, 

 we decided that the marking and recovery re- 

 sults more nearly represent the trapping ratios. 

 This method was used in computing estimates 

 of the number of fry that migrated from above 

 the traps. No marking experiments were con- 

 ducted at Whale Bay and Beartrap Bay. The 

 percentages of fry captured by traps at these 

 two stations were estimated from streamflow 

 studies. 



Adjustments for Incompletely 

 Trapped Streams 



Trapping was discontinued at Whale Bay on 

 May 12. To get a rough estimate of the total 

 migration, we assumed that the fry run there 

 followed a course similar to fry runs in other 

 middle-run streams (Olsen Bay, Fish Bay, 

 Indian Creek, andCannery Creek). The average 

 percentages of the total catch at these four 

 streams for the first part of the fry migration, 

 which extended through May 12, were 26.3 for 

 pink salmon and 33.2 for chum. It seems rea- 

 sonable to assume that the seaward migration 

 of fry from Whale Bay is timed approximately 

 the same; therefore, by extrapolation we be- 

 lieve that if traps had fished throughout the 



Table 2. — Coniparison of streamflow studies and marking — recovery experiments for eight 

 streams in Prince William Sound during the 1957 fry migration 



12 



