878 



Fishery Bulletin 92(4), 1994 



Discussion 



Our analyses indicated that no consistent set of pre- 

 dictors (water temperature, release date, or flow) 

 could explain the travel time of zero-age ocean-type 

 chinook salmon through John Day Reservoir. The 

 predictors for travel time changed each year. The 

 stepwise regression procedure failed to find any sta- 

 tistically significant variables to explain results in 

 1981, and flow was not a statistically significant pre- 

 dictor in any year. However, strong correlations 

 among all predictor variables suggested that flow was 

 nearly equally as likely a predictor as water tem- 

 perature in 1983 and in the combined years, or as 

 release date in 1982 and 1983. 



Release date was included as a predictor variable 

 to provide a generic measure to characterize time- 

 based changes in fish development, such as size or 

 physiological changes that progress over the course 

 of the migration period. Since release date entered 

 the model in two of the three years, this suggested 

 that some time-based biological process may have 

 been important. However, the strong correlations 

 among predictor variables in each year limited the 

 utility of such multivariable regression analyses for 

 identifying the importance of any particular variable. 

 Furthermore, in examining bivariate correlations we 

 found no consistent relationships between migration 

 time and any predictor variable. 



Other measures of migratory behavior should be 

 considered when characterizing the migratory dy- 

 namics of a population. One such measure we con- 

 sidered involved describing the directional 

 intrareservoir movement of fish. We observed that 

 within the body of the reservoir, zero-age ocean-type 

 chinook salmon did not exhibit consistent down- 

 stream movement indicative of a continual, directed 

 seaward migration. The majority of fish that were 

 marked and released at transects throughout the 

 reservoir were recaptured at or upstream from the 

 site of release. This indicated that the population was 

 not consistently displaced downstream passively via 

 current. Based on laboratory observations of coho 

 salmon, O. kisutch, Smith (1982) suggested that 

 smolts in the Columbia River may be oriented mostly 

 head-first upstream during outmigration, thus drift- 

 ing downstream tail-first while being swept seaward. 

 Our results indicate that zero-age chinook salmon 

 do not fit this conceptual model. 



The protracted reservoir-residence times apparent 

 in our data are not necessarily peculiar to Columbia 

 River stocks. Reimers (1973) studied fall chinook 

 salmon in the Sixes River, Oregon, and suggested 

 the optimum size at ocean entry is about 130 mm for 

 that stock. He noted that this length was attained 



by juveniles that remained in fresh or estuarine wa- 

 ters for extended periods of time, suggesting that 

 extended freshwater residence is beneficial to zero- 

 age fall chinook salmon. Extended residence of zero- 

 age chinook salmon was observed in the Columbia 

 River during the late 1950s (Mains and Smith, 1964), 

 and even prior to dam construction (Rich, 1922). 



The absence of a strong relationship between the 

 migration rate and water velocity (flow) for ocean- 

 type chinook salmon contrasts with evidence link- 

 ing travel time to flow (Sims and Ossiander 3 ; Sims 

 et al. 6 ), or developmental (smoltification) state, or 

 both (Giorgi, 1990; Berggren and Filardo, 1993; 

 Beeman et al. 11 ) for migratory yearling stream-type 

 chinook salmon. 



The effects of smolt development on migratory be- 

 havior of zero-age fish are not clear. Zaugg (1982) 

 cited a number of examples that suggested smolt 

 development might be an important process govern- 

 ing migratory behavior of zero-age fall chinook 

 salmon. In contrast, investigations conducted in the 

 Rogue River, Oregon, indicated smolt development 

 was not a requirement for downstream migration in 

 ocean-type juveniles and its importance in affecting 

 the rate of migration was not apparent (Ewing et 

 al., 1980). Although the regression analysis in our 

 investigation used a surrogate variable that may 

 reflect smoltification-related effects (release date), 

 its adequacy in representing such effects has not been 

 verified. Future investigations should include direct 

 assessments of effects associated with developmen- 

 tal processes, such as sodium and potassium ion lev- 

 els and gill ATPase levels, as well as migrant size. 



Berggren and Filardo (1993) also examined the 

 relationship between travel time and a host of pre- 

 dictor variables for zero-age chinook salmon in John 

 Day Reservoir. Their analysis included a subset of 

 our data, as well as similar releases that were ex- 

 ecuted in 1986-88 (Harmon et al. 12 ). In their multi- 

 variable approach, data were pooled across years. The 

 variables in the final multiple regression model in- 

 cluded release date, inverse flow, and an index of the 

 absolute change in flow. The bivariate relationship 

 between smolt travel time and inverse flow had an 

 associated r 2 value of 0.28. In contrast to our results, 

 they concluded that increased flows reduced travel 

 time of zero-age chinook salmon. 



11 Beeman, J. W., D. Rondorf, J. Faler, M. Free, and P. Haner. 

 1990. Assessment of smolt condition for travel time analysis. 

 U.S. Fish Wild. Serv., Cook, WA 98605. Report to Bonneville 

 Power Administration, 71 p. 



12 Harmon, J. R., G. M. Matthews, D. L. Park, and T. E. Ruehle. 

 1989. Evaluation of transportation of juvenile salmonids and 

 related research on the Columbia and Snake Rivers, 1988. 

 Northwest Fish. Sci. Cent., Natl. Mar. Fish. Serv., Seattle, WA 

 98112-2097. Report to U.S. Army Corp of Engineers, 1 1 p. 



