We presume that these fish were attracted up- 

 stream by the cooler water of the Snake River ( 14r- 

 22° C), wliich averaged 2 to 3° C. less than the 

 reservoir. The subsequent downstream migration 

 coincided with an increase in river temperatures to 

 over 18° C. and the breakdown of the convergence 

 line. 



In most years, after the migration of smolts 

 ended in late Jime or July, some chinook salmon 

 remained in the reservoir as holdovers. In 1963, 

 most of the migrants did not leave the reservoir 

 even though the outmigration continued into Au- 

 gust (Sims, 1970). The holdover fish were from all 

 salmon ix>pulations but were mainly progeny of 

 fall cliinook salmon. They were easily distin- 

 guished from more recent arrivals by their large 

 size; in addition, some had identifying fin clips or 

 tags. 



Gill net catches showed that the holdover salm- 

 on had a restricted spatial distribution through 

 the summer and early fall. By late June or early 

 July, surface temperatures exceeded 20° C. and 

 oxygen depletion progressed upward from the bot- 

 tom. These conditions were especially prevalent in 

 the upper end of the reservoir (figs. 6 and 7) . In- 

 creasing epilimnion temperatures and decreasing 

 liypolimnion levels of dissolved oxygen eventually 

 confined the holdovers into restricted areas un- 

 favorable for juvenile salmon (fig. 13). By late 

 September and October, conditions began to im- 

 prove and the surviving juvenile salmon dispersed 

 tliroughout tlie reservoir. 



This same sequence of events occurred each year, 

 but with modification. According to Raleigh and 

 Ebel (1967), the amovmt of reservoir drawdown 

 and the time and duration of the subsequent filling 

 period appeared to be most significant in creating 

 large differences in temperature and oxygen from 

 year to year. In 1965, the large drawdown (28 m.) 

 and prolonged filling (fig. 4) caused late forma- 

 tion of a thermocline, high temperatures, and low 

 oxygen concentrations during late summer. In 

 1964, drawdown was significant (27 m.) but the 

 filling period was shorter. Water temperatures and 

 oxygen concentrations through the summer were 

 more favorable than in 1965 but less favorable 

 than in 1963, when the drawdown was small (3m.) 

 and tlie filling period was early and short. 



Estimates of movement of juvenile fall cliinook 

 salmon into Brownlee Reservoir (Krcma and Ra- 



leigh, 1970) and later escapement (Sims, 1970) 

 verified that the survival of holdover salmon was 

 extremely poor. Fortunately, the conditions that 

 brought about the harshest summer environment 

 (large drawdown and delayed fill) were also the 

 conditions that facilitated rapid passage of finger- 

 lings through the impoundment (Ebel and Koski, 

 1968; Sims, 1970). 



Holdovers of salmon were encountered in the 

 Powder River Arm in October 1962 when 75 chi- 

 nook salmon were caught. Two of these fish bore 

 fin clips that identified them as progeny of fall 

 chinook salmon from the Snake River that had 

 entered the reservoir in the spring. One fish of this 

 group was also recaptured at the upper end of 

 the main reservoir in the spring of 1963. 



In early summer of 1963, fall chinook salmon 

 were captured in gill nets in the lower Powder 

 River Arm. As the environment improved in the 

 fall, many fish moved into the upper arm. From 

 mid-September to mid-December, 561 chinook 

 salmon were captured; of these 444 were tagged 

 and released. The subsequent capture of 28 marked 

 fish in the area of release provided evidence that 

 some fall chinook salmon remained in the upper 

 arm until drawndown in December. On the basis of 

 captures of tagged and untagged fish in the Pow- 

 der River Arm, a population estimate (made by 

 the technique of Schnabel, 1938) of chinook 

 salmon in the arm ranged from 2,631 to 6,521 ; the 

 average of 22 estimates was 3,883. 



The movement of holdovers from Brownlee Res- 

 ervoir in 1963 began in January and peaked in 

 February. A total of 5,396 fish were caught in the 

 skimmer net and in scoop traps below the dam. 

 The exodus in 1964 started in November, peaked in 

 January, and was completed by mid-May (Sims, 

 1970) . The total catch of the skimmer net (partial- 

 ly deactivated in February) and the scoop traps 

 was 1,275 fish. This early outmigration appeared 

 to be a displacement because of the approach of a 

 cold water mass through the reservoir (fig. 14). 



The recapture of marked holdover fall chinook 

 salmon within the reservoir in 1963 provided in- 

 formation on movements of these fingerlings. 

 Early in the year (before April) most of the fish 

 moved toward the dam, but many moved upstream 

 and a few were recaptured in the area of release. 

 The recapture of fish in all areas of the reservoir 

 and the relatively slow rates of movement indi- 



JUVBNILB SALMON DISTRIBUTION AND MOVEMENT IN BROWNLEE RESERVOIR 



235 



