Table 2. — Estimates of density of female pink 

 salmon spawning in three areas in Sashin 

 Creek 



Distribution of the tagged females further 

 demonstrated the higher density in the middle 

 area. The numbers of tagged females spawn- 

 ing in each area woiild be directly propor- 

 tional to the area of spawning bed if the 

 females occurred in equal density. The ex- 

 pected number of tagged females per area, 

 based on the hypothesis that density was 

 equal, is given in table 3 with the observed 

 number. A chi-square test yielded X2(2 d.f.) = 

 5.22 (probability = 0.07). This result com- 

 bined with the fact that observers consistently 

 found greater densities of untagged females 

 in the middle area than in the upper and 

 lower areas causes me to conclude that the 

 density of spawners was not uniform among 

 the three study areas. 



Table 3. — Numbers of tagged female pink salncn 

 observed in three areas in Sashin Creek and 

 the expected number, on the basis of the 

 null hypothesis of equal density 



[X^(2 d.f.) = 5.22 (probability = 0.07)] 



Distribution of Large and Small Females 



Large and snnall females were distributed 

 equally on the spawning ground. Of 86 tagged 

 female pink salmon observed to spawn, half 

 were shorter than average length (49.9 cm.) 

 and half were longer. The hypothesis that 

 large and small females spawned with equal 

 frequency was tested for each study area. 



Table <4.— Nunibers of large and small tagged 

 female pink salmon that spawned in three 

 areas in Sashin Creek, and associated 

 probabilities, assuming the distribution of 

 spawners was unrelated to size 



Table 4 gives the observed nunnbers of large 

 and small fenaales in the study areas and 

 the 2-tailed probabilities of occurrence of 

 the observed values, assuming the hypothesis 

 to be correct. The results clearly demon- 

 strate that distribution was unrelated to length. 



Distribution of Early- and Late-Spawning 

 Females 



I wish to consider two questions on the 

 distribution of early and late spawners: (1) Do 

 large and snnall females spawn early or 

 late with equal frequency? (2) Are early 

 and late spawners distributed similarly among 

 the study areas? Pink salmon began to spawn 

 August 10, and spawning was completed by 

 September 30. About 50 percent of the fe- 

 males spawned before September 3. Forty-four 

 tagged females that spawned before September 

 3 were classified as "early"; 42 that spawned 

 September 3 and later were classified as 

 "late." 



The binomial test was used to examine 

 the first question. Of the 44 early- spawning 

 tagged females, 26 were large and 18 were 

 small. The null hypothesis that large and 

 snnall females spawned early and late with 

 equal frequency was tested with the normal 

 approximation of the binomial distribution. 

 A value z = | 1.06] was obtained. If the hy- 

 pothesis is true, a value z = >|l.06l would be 

 expected in 29 percent of similar experi- 

 ments; I conclude that large and small fe- 

 males tagged August 20 and 21 spawned 

 early or late with equal or nearly equal 

 frequency. 



The chi-square test was used to examine 

 the second question. The expected numbers 

 of tagged females in the study areas (based 

 on the null hypothesis that early- and late- 

 spawning females exhibited no difference in 

 their distribution) and the numbers of early- 

 and late-spawning females observed in each 

 study area are given in table 5. The value 

 X2(2 d.f.) = 7.87 (probability = 0.02) was ob- 

 tained. I conclude that early- spawning females 

 favored the upstream area and late-spawning 



