FECUNDITY OF NORTH AMERICAN SALMONIDAE 



455 



salmon, Oncorhynchu-s keta, in Japanese waters. 

 They point out that chum salmon from Hokkaido 

 rivers have a larfj:er number of eggs tlian chum 

 salmon from rivers in South Kui'ile; however, 

 analysis of their data shows that in both areas 



there is a significant difference between individual 

 rivers. The analysis, based on data from 7 rivers 

 in Hokkaido and 109 specimens of chum salmon, 

 is as follows, using data for the left ovary only to 

 simplify the tabulations: 



F=221,489.2/37,704.7=5.87. For Pof .01, F=2.99 



Similarly for South Kurile rivers, 7^=3.50 with 

 /' for P of .01 of 3.17. For the combined data 

 (243 specimens) which come from 13 rivers, 

 F= 13.74 with F for P of .01 of 2.27. This shows 

 that there is a tendency for each river to have its 



own regression line for egg number on fish length. 

 The fact of differences between the regressions 

 for chum salmon from different rivers results in 

 three regression lines (three center curves of 

 figure 3). The total regression (dotted line) 



>- 



< 

 > 



o 



o 



UJ 

 CD 



1500 



400 



300 



1200 



I 100 



1000 



900 



800 



T 



T 



MEANS 



• - HOKKAIDO 



X - SOUTH KURILE 



53 



54 



55 



56 



57 



58 



59 



60 



62 



63 



FORK LENGTH (CM.) 



FioiRK 3. — Hchitioii of egg iiumbir (left ovary) of Japanese chum salmon, O. kela, to body length to illustrate total 



between-rivers, and within-rivers regressions. 



