FRENCH and DUNN: LOSS FROM HIGH-SEAS GILLNETTING 



I20r 



Short intervols 

 ^ of fishing 



Continuous fishing 



a I20r 



UJ 



CD 



5 100- 



Z 



80- 



60- 



40 - 



20- 



— 



3 6 9 12 



NUMBER OF HOURS FISHED 



2 October 



15 



Short intervals 

 .'•° of fishing 



Continuous fishing 



3 6 9 12 15 



NUMBER OF HOURS FISHED 



Figure 4. — Comparison of salmon catch in gill nets fished 

 continuously with the cumulative catch in nets fished for 

 short intervals (2.0-3.5 h), 1964. Data for short intervals of 

 fishing have been adjusted to allow for comparable fishing 

 hours. 



also retained during the second half of the 

 period. Rewriting C in terms of Ci and Co we 

 have 



C = SiCi + {u-2ll(l)C2 



or 



C/ci = si + {n2lui) (c-ilci). 



By substituting the observed values of C, Cu 

 and C2 for each 6-h fishing period, the regres- 

 sion of (CIci) on (c-ilci) will be linear with 

 intercept equal to .si and slope equal to {H2I111). 



In the application of this model, it is assumed 

 that over the totality of the 6-h-fishing periods, 

 the avei'age number offish available to two units 

 of gear fishing simultaneously is the same. The 

 loss from gill nets, 1 - .si, applies to the frac- 

 tion of fish disappearing from a gill net unit 

 during a 3-h interval which is assumed to begin 

 IV2 h after the fish enter the net. The loss from 

 the time of entry to the beginning of the 3-h 

 interval is not included in the estate of .s-j. 



Salmon catches (sockeye, chum, and pink 

 salmon) by time periods and gear units of the 

 usable experimental dropout sets are listed in 

 Table 3. The total of column ci (the catches in 

 the first 3-h unit) includes appropriate catches 

 of Co where the second 3-h unit also constituted 

 the first unit of the second or third time periods 

 (see Figure 2). Sockeye salmon were the prin- 

 cipal species taken in these sets; species totals 

 were 4,327 sockeye, 353 chum, and 4 pink 

 salmon. 



A plot of the catch ratios, CIci on c'2/ci is 

 shown in Figure 5. In computing the regression 

 constants, each point was weighted according 

 to the weighting function shown in the Figure 

 5. This weighting function arose from two 

 considerations. First, in those instances where 

 the difference in catch between the 6-h unit 

 and the sum of catches of the two 3-h units is 

 large, there is some reason to suspect that the 

 number of fish available to units of gear fishing 

 simultaneously is not the same. Hence, those 

 points should receive less weight in the regres- 



Y = .7325 + 1.0161 X 



Figure 5. — Regression of Cla on colci. C is the catch in 

 the 6-h unit, ti, the catch in the first 3-h unit and c-z. the 

 catch in the second 3-h unit. 



851 



