llJ 



<t llJ 



_I CD 



5i 



(D UJ 

 0|- 



iLl UJ 



q2 

 <UJ 



s^ 



Itl o 



I- «n 



500 



400 



300 - 



200 - 



P'^ 100- 



m 

 bkJ 



OCT. I 

 1961 



JAN. I 

 I 



APR. I 



JULY I 

 1962 



OCT. I 



JAN. I 



APR. I 

 1963 



Figure 2. — Abundance of dead 1961 brood year pink salmon eggs and larvae in Sashin 



Creek study area. 



with the beginning of spawning and a maxi- 

 mum point approximately coincident with the 

 end of hatching. In plotting points for October 

 29, 1962, and March 10, 1963, we subtracted 

 20 eggs per square meter from the mean 

 densities given in table 2 to adjust for presence 

 of 1962 brood year eggs. The curve showing 

 changes in abundance of dead 1961 brood year 

 larvae was drawn by assuming a zero point at 

 the beginning of hatching (a few live larvae 

 were collected in the October 25, 1961, sam- 

 ples). 



Composition of the Population of 

 Dead Eggs and Larvae 



Changes in the composition of the population 

 of dead eggs and larvae in the study area were 

 determined by analyzing changes in the per- 

 centages of (1) whole eggs, (2) egg fragments, 

 and (3) larvae composing the total dead popu- 

 lation. These percentages were calculated in 

 the following manner (illustrated with the 

 percentage of dead whole eggs in k samples): 



k whole eggs 



Percentage of whole eggs = 1 total dead X 100 



Percentages of egg fragments or dead larvae 

 were calculated by substituting one class or 

 the other for whole eggs in the above equation. 



Since samples contained dead specimens in 

 unequal numbers, it was necessary to consider 

 the problem of weighting for binomial varia- 

 tion. Samples containing less than 10 dead 

 specimens were excluded from the analysis 

 to minimize the necessity of applying unequal 

 weights. A method described by Cochran (1943) 

 was used to determine the weighting procedure 

 that would provide the most efficient analysis 

 of the larger samples, that is those containing 

 10 or more dead specimens. Equal weights 

 were found to provide the most efficient 

 analysis and were used in the present study. 



Percentages of dead whole eggs, egg frag- 

 ments, and larvae in samples are given in 

 table 3. These data are plotted in figure 3 for 

 the 1961 brood year. It appeared that dead 

 1961 brood year larvae were no longer present 

 after the first week of August 1962. Also, at 

 about the same time the population of dead 

 1961 brood year eggs comprised mostly frag- 

 ments. In drawing figure 3, we assumed that 

 the observed increase in the percentage of 

 dead whole eggs after August 1962 reported 

 in table 3 was due to mortality of 1962 brood 

 year eggs. 



