FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



239 



Table 1. — Catch of Karluk River red salmon from beginning of the canning industry in 1882 to 1936 



Table 1 gives the yearly catch of Karluk red salmon from the beginning of the 

 commercial fishery in 1882 up to and including the season of 1936. There has been 

 a marked decline in the. abundance of the run of fish. The total runs (catch plus 

 escapement) for the past 16 years ("table 19) have averaged slightly over 2,000,000 

 fish per year, and the average yearly run for 12 of these years was less than 1,600,000 

 fish, whereas for the 7-year period, 1888 to 1894, inclusive, the catch alone averaged 

 over 3,000,000 fish per year. 



In table 2 are presented, for the period 1895 to 1921,* the coefficients of correlation 3 

 between the catches during the years of escapement and the catches 4, 5, and 6 

 years later, together with corresponding values of P.* 



The values of P for 4-year and 6-year intervals are such that the coefficients of 

 correlation cannot be considered statistically different from zero. The value of P 

 for the 5-year interval is such that the coefficient of correlation can be considered 

 statistically significant. It can be concluded from the fact that a statistically signifi- 

 cant correlation of over 0.6 exists between the catches at 5-year intervals and that no 

 statistically significant correlation exists between the catches at 4-year or 6-year 

 intervals that the runs of Karluk red salmon from 1895 to 1921, inclusive, were com- 

 posed largely of 5-year fish. Such a conclusion is verified by the age determinations 

 based on examinations of scale samples taken during 1916, 1917, 1919, and 1921. 



Table 2. 



-Values of coefficients of correlation between catches during year of escapement and catches 

 4, 5, and 6 years later for the -period 1895 to 1921, inclusive 



' Coefficient of correlation. 

 s Transformed coefficient of correlation. 

 > Ratio of z to the standard deviation of z. 

 1 Probability that z is not different from zero. 



! The data for the years 1895 to 1921 were used in this analysis as the fishing effort was fairly constant during this period. 



' Where the relationship between two variables is found or assumed to be linear, the coefficient of correlation r measures the 

 proportion of the variation in one variable which is associated with the second variable. As the number of pairs of observations are 

 relatively small the method of analysis given by Fisher (1930, p. 163) was used. 



' P is the probability that the value of the transformed coefficient of correlation z would have been obtained by chance, i. e., a 

 value of P of 0.01 indicates that if the true value of z was 0.0 a value as large as the one obtained would occur only once in 100 random 

 samples. The relationship between z and r is such that the values of P also indicate the statistical reliability of r. 



