in the comparison because of insufficient tag 

 recoveries. This high value of P indicates a lack of 

 segregation of lateral stream spawning populations 

 by time of occurrence in the run. 



The comparison between the total recovery dis- 

 tributions of Naknek, Brooks, Grosvenor, and 

 Coville Lakes (table 5) yielded a P of less than 

 0.005, suggesting a difference in time occurrence in 

 the run for the populations destined for these 

 lakes. However, almost two-thirds of the chi- 

 square value of 56.57, d.f. = 18 is due to the occur- 

 rence of a greater than expected proportion of 

 early-run tagged fish in the Brooks Lake escape- 

 ment (fig. 7). ff the Brooks Lake tag recoveries 

 are excluded from consideration and a comparison 

 made only between Naknek, Grosvenor, and 

 Coville Lakes recovery distributions, the resulting 

 X 2 =7.43, d.f. = 12, P=0.82. Thus, the recovery 

 distributions for these three lakes are not sig- 

 nificantly different, and segregation of their 

 spawning populations as a whole is minimal. For 

 Brooks Lake, however, results suggest that it 

 derived a relatively greater proportion of its 

 spawning population from the early part of the 

 run than did the other three lakes. 



Table 5. — Numbers of tags recovered by dates of tagging for 

 Naknek, Brooks, Grosvenor, and Coville Lakes, 1962 



Comparison of the tag recovery distributions of 

 unlike spawning grounds, which include the 

 grouped lateral streams of all lakes, Brooks and 

 Grosvenor Rivers (representing interconnecting 

 streams), and American and Hardscrabble Creeks 

 (representing major valley streams) (table 6), 

 yielded a value of x 2 = 14.87, d.f. = 8, P=0.06. 

 This value is essentially at the level of probability 

 (P=0.05) below which values of P are considered 

 to indicate unlike tag recovery distributions. The 

 cause of the large x 2 and resulting low probability 

 level is associated with the absence of early-run 

 (June 24 to July 2) tagged fish in the Grosvenor 

 River recoveries (fig. 6). More than one-third of 



the x 2 value of 14.87 is caused by the lack of these 

 early-run tagged fish. Although, statistically, the 

 test does not indicate a tag recovery distribution 

 for Grosvenor River unlike that of the other 

 spawning grounds, the lack of early-run tagged 

 fish in the recoveries and the resulting large chi- 

 square suggest that this area may have derived its 

 escapement primarily from the latter part of the 

 run. If Grosvenor River recoveries are excluded 

 from consideration and a comparison made only 

 between the grouped lateral streams, American 

 Creek, Brooks River, and Hardscrabble Creek, the 

 resulting x 2 =8.70, d.f. = 6, P=0.20. 



Table 6. — Numbers of lags recovered by dales of lagging for 

 unlike spawning grounds, Naknek River system, 1962 



Discussion 



There is some evidence of heterogeneity between 

 the individual spawning ground tag recovery dis- 

 tributions as compared above. Only Brooks Lake 

 had a tag recovery distribution that differed signifi- 

 cantly from those of the other spawning grounds. 

 The distributions for both the lateral streams and 

 the escapement as a whole show that a higher 

 proportion of early-run fish entered Brooks Lake 

 than the other three lakes. Grosvenor River re- 

 coveries indicate that it may have received its 

 escapement primarily from the middle and later 

 parts of the run (i.e., after July 2). 



Apart from the above situations, however, the 

 evidence presented fails to indicate any clear 

 segregation of most of the known spawning popu- 

 lations composing the Naknek run. It is clear that 

 the recovery distributions for most spawning 

 areas are not significantly different from one 

 another or from the seasonal escapement pattern 

 in the trunk stream. Most areas apparently 

 derive their spawning populations from all parts 

 of the run and, generally, in proportion to the 

 daily escapement. 



472 



U.S. FISH AND WILDLIFE SERVICE 



