So 



"2- 



z « < 



= S0 



* u J 



O Q — 



<0 



047 



r= 0.022 

 p,0.05 

 (>ol iignificoni 



3 4 6 8 10 12 14 16 



TOTAL NUMBER OF FISH RELEASED AT SPRING CREEK (MILLIONS) 



Figure 3.--Correlation between total release (fry and 

 fingerlings, in numbers of fish) and adjusted return 

 of females as defined in appendix table 9, by brood 

 year (1941-49). 



So 

 oft 



2 ffl < 

 § ^ D 



lij q; X 

 « U t 



So" 



Q «^ 



< o 



4 6 8 10 12 14 16 



TOTAL POUNDS OF FISH RELEASED AT SPRING CREEK {THOUSANDS! 



that the survival rate for fingerling releases 

 is many times greater than the survival rate 

 for fry releases. 



Variations of this type, however, may be 

 studied more effectively by making use of 

 multiple regression techniques. "Adjusted re- 

 turns" will refer to the number of females 

 returning 4 years after the releases for the 

 specified brood year, adjusted for the effects 

 of the gill net fishery below Bonneville Dam. 

 The following notation will be used: 



X = pounds of fingerlings released at Spring 

 Creek Hatchery (range" from zero in 

 1942 and 1948 to 13,155 in 1947) 



X = pounds of fry released at Spring Creek 

 Hatchery (range from 2,765 in 1945 to 

 13,239 in 1949) 



y = adjusted returns (range from 1,992 

 for the 1941 brood to 7,846 for the 

 1945 brood) 



Figure 4.--Correlation between total release (fry and 

 fingerlings, in pounds) and adjusted return of females 

 as defined in appendix table 9, by brood year ( 1941-49). 



ANALYSIS OF DATA 



Regression Studies, 1941-49 Brood Years 



In the process of defining the variables to be 

 studied some indication of the relative survival 

 of fry and fingerling releases is obtained. For 

 this purjxjse, deviations from the regression 

 line in figure 4 may be compared with the 

 relative pounds of fry and fingerling releases 

 at Spring Creek Hatchery (appendix table 5). 

 Points below the regression line are all asso- 

 ciated with high percentages of fry releases 

 (89, 100, 100, 73 percent fry by pounds for 

 brood years 1941, 1942, 1948, 1949, respec- 

 tively), whereas the points above the line are 

 associated with much smaller percentages of 

 fry (67, 55, 25, 25, 22 percent for brood years 

 1943, 1944, 1945, 1946, 1947, respectively). 

 This result indicates that the return per pound 

 of fingerlings released is somewhat higher than 

 the return per pound of fry released. Generally, 

 a pound of fry contains 5 to 10 times as many 

 fish as a pound of fingerlings, so it appears 



The multiple correlation coefficient (r 



y.x^x^ 



is 0.891 (P<0.01). The direct correlation be- 

 tween pounds of fingerlings released and return 

 (r ) is 0.890 (see fig. 5a), so that the partial 



correlation between fry releases and return is 

 very close to zero. Insofar as the ranges of 

 X and X are large and ofthe same magnitude, 



the resulting correlations indicate that varia- 

 tion in returns is controlled almost entirely 

 by fingerling releases. This is in agreement 

 with the results of the Washington Department 

 of Fisheries study previously mentioned. 



Incomplete results of a marking experi- 

 ment" which was initiated on the 1956 brood 

 at Spring Creek Hatchery also show a much 

 greater survival rate for fingerling releases. 

 In this experiment equal numbers o1^ two groups 

 of unfed fry were marked by removing the 

 adipose and left pectoral fins and the adipose 



" Bureau of Commercial Fisheries, experiment now 

 in progress. Preliminary tabulations from marked re- 

 turns reported by Oregon Fish Commission, Washington 

 Department of Fisheries, and Bureau of Commercial 

 Fisheries. 



