BARKLEY: SELECTIVITY OF TOWED-NET SAMPLERS 



1 Mesh losses probably significant, analysis with avoidance theory inappropriate. 



2 Mesh losses may be significant, analysis with avoidance theory questionable. 



3 Sample size inadequate, results of analysis may be seriously in error. 



cate 6-hr tows were taken every night for (usu- 

 ally) five successive nights at each of five lo- 

 cations (lat 12°N, 7.5°N, 3.5°N, 0°, and 3.5°S 

 at long 145°W) during these cruises. Catches 

 of skipjack larvae and juveniles during each 

 of these cruises are summarized in Table 2. 



Length-frequency data from three cruises, 

 numbers 43, 44, and 48, were kindly made avail- 

 able to me by Walter M. Matsumoto. These data 

 were similar for all three cruises and were there- 

 fore pooled to obtain a smoother curve (Figure 8, 

 right panel) . Of the 510 animals in this sample, 

 6 could not be measured. Another 5, the largest 

 (see Figure 8, right panel) were suspected of 

 being atypical, possibly moribund and unable to 

 properly avoid the trawl. Figure 8 (left panel) 

 shows the data for the remaining animals, con- 

 verted to speed frequency using Ue = lOL and 

 U = 150 cm/sec. 



Since the skipjack population structure as a 

 function of length, A^l, is completely unknown, 

 two extreme assumptions are considered: First, 

 that avoidance has no effect, so that virtually all 

 of the drop in catch with increasing length (Fig- 

 ure 8) is due to changes in the population. Sec- 

 ond, that population structure has no effect, so 

 that the length-frequency curve is determined 

 only by avoidance and mesh losses. 



The first assumption is illustrated in Figure 9, 

 left panel, where the lowest curve shows speed- 



frequency values. The straight line represents 

 Nl, assuming the largest possible exponential 

 decrease in population with size, consistent with 

 the catch data. The upper curve shows Pc, ob- 

 tained by dividing the catch data by Nl after 

 setting No = 1.0, exactly as was done on Fig- 

 ure 4. This analysis results in a wide range of 

 values for Pc and Xo/R. No in this case becomes 

 100%, equivalent to 500 individuals per class 

 interval. Figure 9, right panel, shows Pc plotted 

 on the linear theoretical graph of Figure 3. 



The alternative extreme, assuming that Nl = 

 No so that population has no influence on catch 

 length frequencies, is illustrated in Figure 10. 

 There is no single apparent "best fit," between 

 observation and theory, so three alternatives are 

 considered. The upper curves, marked A, show 



Table 2. — Skipjack larvae and juveniles caught with 

 the Cobb pelagic trawl during RV Townsend Cromwell 

 equatorial cruise series. Samples were taken along long 

 145°W, at lat 12°N, 7.5°N, 3.5°N, 0°, and 3.5°S. 



Cruise 

 no. 



Dates 



Number 

 of tows 



Number of 



successful 



tows 



Skipjack 

 caught 



813 



