(less than 5 feet deep), most tag color combina- 

 tions were easily identified, but in deep-water 

 spawning areas (more than 5 feet deep), most 

 observers reported some difficulty in positively 

 identifying green tags and in distinguishing light 

 blue and yellow tags from white. Red tags were 

 identified the most easily in deep water. Thus, 

 the assumption of equal likelihood of observation 

 for all tag color combinations would not hold 

 true for fish in deep water. The degree of effi- 

 ciency of observing tagged fish in deep water 

 depends on the depth of the water and the tag 

 color combination. Fortunately, most known 

 spawning in the Naknek system takes place- 

 in water that is less than 5 feet deep. 



In analyzing recoveries for the experiment, I 

 compared the following spawning areas and the 

 escapement in the manner described above. 



1. Individual lakes, i.e., Naknek, Brooks, 

 Grosvenor, and Coville. In this comparison 

 the corrected upstream Brooks River weir count 

 of tagged fish was used as the total recoveries for 

 the lake. This was done because few tags were 

 returned from the Brooks Lake tributary streams. 

 The upstream tag count was corrected by de- 

 ducting tagged fish that moved back downstream 

 through the weir to spawn in Brooks River. 

 For the other lakes, tag recaptures and sightings 

 obtained by all methods (i.e., during stream 

 surveys, at spawning stream weirs, by beach 

 seining, and by skin diving) were totaled for all 

 streams draining into each lake. 



2. Individual, large major valley or terminal 

 streams with similar physical characteristics and 

 periods of spawning activity and with recorded 

 escapements in excess of 10,000 fish. These 

 include American, Hardscrabble, Margot, and 

 Bay of Islands Creeks which drain into Coville, 

 Grosvenor, and Naknek Lakes respectively (fig. 

 2). Headwater Creek, major tributary of Brooks 

 Lake, had too few recoveries for comparison 

 (table 2). 



3. Interconnecting streams (Brooks and Gros- 

 venor Rivers). 



4. Grouped small lateral tributary streams of 

 all four lakes combined. 



5. Naknek River area at the outlet of Naknek 

 Lake. 



Recoveries for days or periods of tagging were 

 arranged in contingency tables for comparisons 

 between selected spawning areas, and the pro- 



portion of tags in each period was tested for 

 independence by chi-square. Spawning areas 

 compared in this manner were essentially those 

 listed above. In addition, tag recoveries from 

 the small lateral streams of each lake were grouped 

 to provide adequate numbers for testing and were 

 treated as a single spawning unit. Tag recovery 

 proportions for lateral streams of each lake were 

 compared and tested. Comparisons and tests 

 were also made between spawning areas of unlike 

 physical characteristics, escapement records, and 

 spawning periods (i.e., between the large major 

 valley streams, small lateral streams, and inter- 

 connecting streams). 



For most of the areas compared, tag recoveries 

 from single days of tagging were too few to permit 

 tests of independence on a day-to-day basis, 

 and sufficient recoveries were obtained by com- 

 bining the recoveries from several continuous 

 days of tagging into a number of periods that 

 included all 23 days of tagging. Tests of inde- 

 pendence were thus made on a period-to-period 

 basis. 



Tests for segregation were grouped into tagging 

 periods based on the area of spawning. In the 

 lakes, the 23-day tagging experiment was divided 

 into seven tagging periods with 3 days in each 

 period, except the first and last periods, which 

 had 4 days. In the other spawning areas, it was 

 necessary to group the laggings into three periods. 

 These periods were unequal in length because they 

 were based on the daily escapement pattern ,of 

 the run (table 1). The first period, representing 

 early-run salmon, included recoveries from laggings 

 between June 24 and July 2. This period con- 

 tained an early peak of abundance, which is usually 

 characteristic of the Naknek sockeye salmon run. 

 This early peak may represent spawning groups 

 that are segregated in time of occurrence and 

 destined for particular spawning areas. The 

 second period representing middle-run fish, which 

 compose most of the escapement, extended from 

 July 3 to 8, inclusive. The third period, repre- 

 senting "tail-of-the-run" fish, included recoveries 

 from July 9 to 18 taggings. 



Age Segregation by Time of Occurrence on the Run 

 and on the Spawning Grounds 



Segregation of population by age groups by time of 

 occurrence in the run was studied as follows. The 

 percentages of the various age groups in the daily 



468 



U.S. FISH AND WILDLIFE SERVICE 



