Table 5. — Conparison by tagging experiment of recoveries of marked red salmon at principal recovery points, 



Kviohak River system 



^ Grouped because of small returns from experiment 4. 



Statistical Comparison for Iliamna 

 Lake and Lake Clark 



All Iliamna Lake recoveries combined and 

 all recoveries in the Newhalen and above, 

 also combined, are arranged by experiment 

 in table 6 where they are headed simply, 

 "Iliamna Lake" and "Lake Clark." Chi-square 

 tests of the hypothesis of like recovery distri- 

 butions in the two lakes yielded values p= 0.008 

 for 1957, p = 0.700 for 1958, and p - 0.613 for 

 1959 data. Clearly the 1957 distributions 

 differed widely; 1958 and 1959 distributions 

 were strikingly similar. It is apparent that 

 only in 1957 was there marked segregation; 

 more of the early salmon went to Lake Clark. 

 It is now important to determine whether the 

 segregation applied to the entire system above 

 Iliamna Lake or merely to parts of it. This 

 may be resolved by a process of elimination. 



In figure 5 distributions at recovery points 

 25, 26, 27, and 28 involved small numbers of 

 tags, apparently distributed similarly to the 

 composite. These are not responsible for the 

 low level of probability associated with the 

 1957 test. Newhalen-Nondalton recoveries have 



already been compared with important re- 

 covery points on Iliamna Lake. They were 

 accepted as distributed in the same propor- 

 tions, though with some reservation because 

 the visual comparison showed contrast. The 

 combined Gibraltar, Copper, and Knutson 

 Complex recoveries were distributed in the 

 same proportion as the remainder of Iliamna 

 Lake recoveries grouped (p := 0.806), so no 

 discrepancy was introduced by lumping all 

 Iliamna Lake data. The cause of the low 

 probability level associated with the test of 

 Iliamna Lake data against all data from above 

 Iliamna Lake appears then to be Kijik Lake. 



In summary, no significant difference is 

 apparent in the distributions to Iliamna Lake 

 and Lake Clark in 2 of 3 years. The difference 

 in 1957 appears positively associated with the 

 Kijik spawning system. Kijik itself, while con- 

 trasting with the remainder of the system, is 

 not clearly segregated. Its escapement data 

 exhibit somewhat independent but also in- 

 consistent timing and form, since they differed 

 from the composite in a different way each 

 year. 



18 



