Figure 5.--Recoveries by experiment, 1957, Kvichak River system. The combined recoveries from all experi- 

 ments are plotted as a bar graph in the box labeled "seasonal composite." Recovery distributions at the indi- 

 vidual points are plotted as bar graphs and are shown at the appropriate locations on the map (numbers on 

 abscissa are from tables 1, 2, or 3). Recovery distributions to be compared statistically are identified by a 

 letter above the graph that corresponds to the name in the sequence on the lefthand margin. (See figure 8 

 for locations of recovery points.) 



Statistical Comparison for Principal 

 Recovery Points 



The statistical tests are an objective ap- 

 proach to showing whether segregation does 

 or does not exist. It is particularly important 

 that numerically large spawning groups be 

 tested in this way, for they can play an im- 

 portant part in the management of the fishery 

 if they are segregated. 



Recoveries from each of the important re- 

 covery points are arranged in table 5 by ex- 

 periment, except for experiments Sand 4 which 

 are combined in 1958. In that year experiment 

 4 was not sufficiently represented for a test 

 in the preferred manner. 



Chi-square tests of the hypothesis of like 

 recovery distributions among four recovery 

 points in 1957 and five recovery points in 

 each of 1958 and 1959 gave values p= 0.203, 

 P = 0.400, and p = 0.929 respectively. It is 

 significant that the high levels of p in 1958 

 and 1959 are associated with the strongest 

 tests, i.e., more areas tested and greater 

 numbers of recoveries. Since the values of p 

 are high in all 3 years, segregation in time 

 of passage at Igiugig is probably minimal. 

 It is important, however, to consider whether 

 these recoveries are representative of the 

 single spawning groups (as defined) utilizing 

 specific spawning grounds. 



On Copper and Iliamna Rivers the recoveries 

 are surely representative of those systems. 



15 



