Table 13.— Time difference (in days) between 

 average weekly catches of 10.000 and 20,000 

 fish in traps in lower Chatham Strait and two 

 inner areas and miles traveled per day (M/D) 

 calculated from the time difference. 



-'Twenty thousand fish per week not reached. 



and migration out of the fishing area, either into 

 spawning streams or into closed areas. Tagged fish 

 were also exposed to possible immediate mortality 

 from the tagging process, continuing mortality caused 

 by the attached tag, and a spurious mortality through 

 loss of the tag. Nonreporting of tags by the fishermen 

 decreases the apparent fishing mortality or exploita- 

 tion rate, but if the fraction of recoveries that are un- 

 reported is constant during all recovery periods, non- 

 reporting of tags does not affect the estimate of the 

 total survival rate when the estimation method 

 described by Paulik (1962) is used. 



Only trap recoveries were used in the survival rate 

 computations. All recoveries, regardless of the type of 

 recapture gear, were used to estimate the exploitation 

 rates. Although the seines took about 30% of the 

 catch, only 16% of the tag recoveries were made by 

 seiners. It is possible that the seine fleet may have 

 been operating on stocks not adequately sampled by 

 the trap-based tagging operations or that the recovery 

 of tags from seine catches was less efficient than from 

 trap catches. 



The hypothesis that the mean days out for seines 

 was less than or equal to the mean days out for traps 



was tested for each release for which 10 or more seine 

 recaptures with known recapture dates were 

 available. If the trap and seine variances did not differ 

 significantly, the usual t-test was used; if the 

 variances were significantly different, the modified t- 

 test proposed by Welch (1938) was used. The i-values 

 obtained are listed in Table 15. The results of the in- 

 dividual tests were combined by means of Fisher's 

 chi-square statistic (Fisher 1950). The overall conclu- 

 sion based on the results of Fisher's test is that the 

 average days out for the seine recoveries is significant- 

 ly greater than the average days out for trap 

 recoveries. It appeared that in many cases the seines 

 were more likely to be the last gear encountered by 

 salmon before entering the spawning streams. Since 

 the mean number of days out has an inverse 

 relationship to the negative natural logarithm of the 

 survival rate, this difference between the gears would 

 be expected to affect the survival rate estimates. For 

 this reason the survival rate estimates were computed 

 from the trap recoveries only. 



Excluding seine recoveries when the average days 

 out is greater than for traps could introduce a negative 

 bias in the estimates of survival rates. That is, our es- 

 timates would usually be slightly more if we had in- 

 cluded them. However, the difference in days out was 

 usually not great enough to introduce an important 

 bias, and considering the various inadequacies of the 

 seine recoveries, we do not believe their inclusion 

 would improve the conclusions of this report. 



The information on days out for the trap recoveries, 

 although considered to be more accurate than for the 

 seine recoveries, is also suspect. During much of the 

 season the traps were not lifted every day. 

 Nevertheless the original tag recovery records showed 

 an exact date of recapture for nearly every recovery. 

 For a representative sample of experiments, recapture 

 dates for all tag recoveries made in a particular trap 

 were compared to the recorded lifts of that trap as 

 reported in the daily catch records. A recovery 

 reported on a day when no lift was made was 

 tabulated as a disagreement. It is obvious that this 

 analysis yields only a minimum estimate of the actual 

 misclassification because if daily lifts were made, no 

 misclassifications could be detected. The percent dis- 

 agreement per tagging experiment ranged from 7% to 

 48%. For most releases tag recaptures were reported 

 during the weekend closures. Because of this factor 

 and the difficulty in determining the exact day of 

 recapture, weekly time periods were used to estimate 

 survival rates. The number of weekly recovery periods 

 was approximated to the nearest week. 



Weekly survival rates were calculated for 30 groups 

 of tagged fish for which the total recovery period ex- 

 tended two or more weeks. None of the groups tagged 

 in 1940 and 1945 met this criterion. It was assumed 

 that the weekly survival rate remained constant for 

 each group from the time of release until the end of 

 the fishing season. The computational method used 



30 



