April through May and Columbia River fall chinook salmon 

 releases. The correlation coefficient for the regression of annual 

 Washington ocean sport catch in 1964-75 on total Columbia 

 River fall chinook salmon releases was 0.958. The comparison 

 of actual catches with those predicted from this regression is il- 

 lustrated graphically in Figure 4. 



260-, 



240- 



| ACTUAL 

 □ PREDICTED 



Average error is 6.2% 



X 200- 



Figure 4. — Annual Washington ocean sport catch of chinook salmon, 1964-75, and 

 that predicted from the regression of catch on total Columbia River hatchery 

 releases of fall chinook salmon of brood year groups /'-3 and i-4 combined. 

 Numbers at the tops of the columns indicate percentage error. 



Previous fishing success. — Troll catch and CPUE in most 

 areas were significantly correlated with 1) troll CPUE during 

 September through October of year /— 1 in the Coastal 

 Washington district, 2) troll CPUE in Canadian areas 24-27 

 during September through October of year /— 1, 3) annual 

 Puget Sound sport catch in year i—l, and 4) Canadian purse 

 seine catches of jacks during August of year i—l in Canadian 

 area 20. The correlations with Puget Sound sport catch may be 

 serial correlations, since this variable was highly correlated with 

 Columbia River fall hatchery releases. Although the area 20 

 purse seine catches were also highly associated with Columbia 

 River hatchery releases, they may be of predictive value as an in- 

 dicator of hatchery plant survival. For example, Figure 5 il- 

 lustrates the relationship between early season Canadian troll 

 catches and the natural log of numbers of jacks caught. Correla- 

 tion coefficients for the regressions of 1) annual Washington 

 ocean sport catch and 2) La Push-Neah Bay sport catch during 

 April through October in 1964-75 on logs of jack catches were 

 0.887 and 0.913, respectively. 



I 



180-1 



160- 



140- 



120- 



U- 



o 



c/) 100- 



a 



z 

 < 



<£ 80- 

 O 



60- 



40- 



QI974 



1973 O, 



O 1972 

 QI975 



~ ,-J966 



oi97i q-) 



1970 



O 1967 

 1968 O OI969 



QI964 



O 1965 



QI963 



r=937 



~~ l — 

 3.2 



3.4 



3.6 



— T" 

 3.8 



4.0 



4.2 



4.4 



LN NUMBER OF JACKS 



Figure 5. — Canadian commercial troll catch of chinook salmon in areas C, 21. and 

 23 during April through May vs. natural log of number of jacks in the area 20 purse 

 seine catch during August of year /"- 1, 1963-75. 



Prediction of ocean catches.— A simple linear regression with 

 lower Columbia River hatchery releases of fall chinook salmon 

 of brood year groups ;'-3 and i-4 combined as the indepen- 

 dent variable was the best predictive method. 



Coho Salmon 



1. Troll catches were significantly correlated with levels of 

 nominal fishing effort. 



2. There were no significant associations between ocean sport 

 catches and the number of angler trips. 



3. Troll and sport catches were significantly correlated with in- 

 dices of brood year abundance of late-run Columbia River 

 coho salmon in Ricker curves. 



4. Troll catches off southern and central Washington were 

 significantly correlated with Columbia River jack returns 

 the previous year. 



5. Ocean fishing success was not significantly correlated with 

 indices of hatchery production, either in total or on an 

 area-by-area basis. 



Levels of effort. — Troll catches were significantly correlated 

 with levels of troll effort during 1955-75, as indicated by the cor- 

 relation coefficients listed below. In contrast, we found no 

 significant associations between sport catches and the number of 

 angler trips. 



Area 



June-August 



Columbia River 



0.800 



Coastal Washington 



0.880 



Puget Sound (Washington) 



0.603 



Areas C, 21, 23 (Canadian) 



0.865 



Areas 24-27 (Canadian) 



0.802 



