126 



FISHERY BULLETIN OF THE FISH AND WILDLIFE SERVICE 



over the period from 1912 to 1937, with the exception of occasional years for which 

 no figures were available. It is unfortunate that similar data are not available for 

 the entire river. 



During this long period the catch delivered to the association has averaged 

 nearly 25 percent of the total deliveries on the Columbia River, and has ranged 

 quite consistently between 20 and 30 percent. To test the reliability of these data 

 as an index of changes in relative abundance during different periods, the Pear- 

 sonian coefficient of correlation, "r," has been calculated between the total annual 

 deliveries to the company and the total deliveries for the entire fishery as given in 

 the report by the Oregon State Planning Board (1938). Between 1912 and 1937 

 there were 20 years for which complete records were available, and for these the 

 coefficient of correlation is 0.86. The records appear to show, however, that some 

 change took place about 1934, so that the records for the last 3 or 4 years are not 

 consistent with those for earlier years. We have, therefore, calculated "r" for the 

 16 years of record between 1912 and 1928. The value is practically 0.9. Both show 

 such a high degree of correlation that reasonable confidence may be placed in the 

 assumption that the deliveries to the Columbia River Packers Association will serve 

 to indicate long-time (secular) changes in relative abundance of chinook salmon in 

 different parts of the season. 



Table 16. — Monthly totals of deliveries of chinook salmon to the Columbia River Packers Association, 

 1912-37, in thousands of pounds, for the spring fishing season only 



Notb. — The years 1929, 1930, and 1935 are omitted because of incomplete records. 



From data given in table 16 we have calculated the trends by the method c! 

 averages, and these are shown in figure 4, which has been put on a semilogarithmic 

 grid so that relative changes will be correctly shown and the several trends can be 

 directly compared. It is apparent from this that while a general reduction has 

 taken place, as is shown in each month and also in' the total, the reduction in the 

 July catch has been by far the greatest. From a value of nearly 2,000,000 pounds 

 at the beginning of this period (1912), the line of trend of the July deliveries has 

 dropped to only about 600,000 pounds in 1937. The present deliveries are, there- 

 fore, approximately one-third of what they were during July 25 years ago. At the 

 same time the totals for the entire spring fishing season have dropped from about 

 6,000,000 pounds to about 3,500,000 pounds. This graph also shows that the deliv- 

 eries during May have been seriously reduced. Curiously enough, the trend of the 

 June deliveries is approximately the same as for the spring season as a whole, although 

 those of May and July show evidence of much more serious depletion. Deliveries in 

 August have not suffered nearly so much as those of the other months of the spring 

 season — perhaps because of increased utilization of these later running fish which arc 

 not of so good a quality as those of May, June, and July. 



