Fish and Wildlife Service, 1960, vol. II, p. H- 2) 

 for evidence of such an effect, and it was con- 

 cluded that Spring Creek Hatchery fish probably 

 contribute to the fishery in proportion to their 

 abundance. Preliminary data on 1959 returns 

 of 3-year-old marked Spring Creek Hatchery 

 releases (adipose-right pectoral) gives con- 

 firming evidence, with approximately 35 per- 

 cent of the marked fish recovered in the river 

 taken by the fishery' and 30 percent of the 

 total run taken by the fishery'. Returns of 

 marked chinooks in 1960, 1961, and 1962 should 

 give a more precise evaluation. 



Variables Concerned with Hatchery Releases 



For many years biologists have attempted 

 to determine when juvenile chinook salmon 

 should be released in order to yield the best 

 survival to adulthood. Experimentation includ- 

 ing the release of marked fish has led to con- 

 tradictory results, and opinions on this subject 

 are widely divergent. In a multiple regression 

 study by Washington Department of Fisheries " 

 on the tagged adult returns at two hatcheries 

 against releases at ages of 30 days or less, 

 60 days, 90 days, etc., very high multiple 

 correlation coefficients were obtained. At both 

 hatcheries it was indicated that releases in 

 the vicinity of 90 days were very much more 

 productive than any others, and releases before 

 30 days gave no significant indication of pro- 

 duction (partial correlation coefficients very 

 close to zero). 



A similar study of Spring Creek Hatchery 

 data would have been desirable, but major 

 changes in hatchery facilities in 1950 and in 

 1953 made it necessary to divide the data into 

 three parts, the brood years 1941-49, 1950-52, 

 and 1953-55. As a result, only the first period 

 (1941-49) contributed sufficient data for any 



Estimated from recoveries in the Columbia River 

 gill net fishery, 30 percent of which was sampled by 

 Oregon Fish Commission and Washington Department of 

 Fisheries, and from hatchery recoveries, 



' Catch plus Bonneville count was used as the total 

 run as in appendix table 7, 



1" Washington Department of Fisheries, unpublished 

 manuscript, 1958. 



correlation study, and, with only 9 years of 

 data, no more than two independent variables, 

 fry and fingerling releases, could be used. The 

 fry size (500 per pound) is comparable to the 

 size of 30-day-old chinook - the apparent age 

 below which no significant production was in- 

 dicated in the Washington Department of Fish- 

 eries study. 



For many of the Spring Creek Hatchery re- 

 leases in the 1940's no date of release was 

 given, so it was necessary to classify by 

 average weight rather than by age at release. 

 This classification rules out theuseof the 1939 

 and 1940 brood years. In 1939 no weights or 

 time of release were given. In 1940 only the 

 total weight for all fish released was given, 

 and this weight included large numbers re- 

 leased in Big White Salmon River and large 

 numbers released over a 2-month period at 

 Spring Creek Hatchery without specified dates 

 or numbers for the individual releases. 



Since only two groups (fry and fingerlings) 

 may be used in a regression study and since 

 each group encompasses a wide range of sizes, 

 there is some advantage in using pounds of 

 fingerlings (or fry) released as the independent 

 variable instead of number released. If the 

 number released is used, the inherent as- 

 sumption that eventual survival is independent 

 of size within each group (fry or fingerlings) 

 may inhibit any correlation with returns. On 

 the other hand, the use of pounds released is 

 equivalent to a weighting process which as- 

 sumes that within each group survival is pro- 

 portional to the weight of a fish at the time of 

 release, and the correlation with returns may 

 be improved. That this is actually the case can 

 be seen when adjusted female returns to Spring 

 Creek Hatchery are plotted against total num- 

 bers (fry and fingerlings) released in Spring 

 Creek Hatchery and total pounds released 

 (figs. 3 and 4), with respective correlation 

 coefficients of 0.022 (not significant) and 

 0.682 (P<0,05). Insofar as economic factors 

 are to be considered, the use of poundage is 

 obviously more meaningful since costs are 

 more closely related to pounds of fish released 

 than numbers. 



