FISHERY BULLETIN: VOL. 70, NO. 3 



% OF TOTAL 

 CATCH 



10.0 

 8.0 



6.0 

 4.0 



^0 



1.0 

 08 



0.6 

 0.4 



02 



NUMBER 

 CAUGHT 



% OF TOTAL 

 CATCH 



50 60 

 40 



30 



20 



10 



50 



40 



30 



20 



10 



II 



N-504 



10 



lb n r-. Ln 



12 



0.1 



0.2 



0.3 



10 



20 



"e/u 



30 40 



LENGTH (mm) 



50 



60 



70 



Figure 8. — Speed-frequency curve for skipjack larvae caught by a Cobb pelagic trawl (left panel) ob- 

 tained from a length-frequency curve (right panel). Data courtesy of W. M. Matsumoto, NMFS, South- 

 west Fisheries Center, Honolulu Laboratory. 



the data aligned to yield a fit for the best sam- 

 pled class intervals. The lower curve, C, is 

 aligned so as to fit the theoretical curves in re- 

 gions of maximum slope, where sampling is most 

 sensitive to avoidance. The middle curve, B, is 

 a compromise which yields a reasonable fit at 

 intermediate class intervals. Although all three 

 analyses look plausible in the left panel of Fig- 

 ure 10, the linear graph shown on the right of 

 Figure 10 clearly demonstrates that curve B 

 yields the best match between theory and ob- 

 servation. 



In principle there is no reason to prefer the 

 analyses shown on Figure 10 to their alternative, 

 Figure 9. However, I personally regard curve B 

 on Figure 10 as the best solution because it pro- 

 vides the simplest and most informative expla- 

 nation for the observations. 



Note that all three sets of curves on Figure 10 

 fit two separate theoretical curves, in two dis- 

 tinct and well-separated groups of class inter- 

 vals. As in previous examples, Pc for the largest 

 animals markedly exceeds the values which 

 would be expected if the theoretical Xo/R curve 

 for smaller animals were simply extrapolated 

 toward the right (except for curve A). In the 



cases of Stolephorus and anchovy, these anoma- 

 lies were relatively minor and could reasonably 

 be attributed to selective sampling of animals 

 with subnormal avoidance ability. In the case 

 of skipjack, however, this anomaly is more pro- 

 nounced. 



A more likely explanation for the two-part 

 curves of Figure 10 is the construction of the 

 Cobb trawl, which is lined with netting of two 

 different mesh sizes: 19 mm stretched mesh 

 of 9-thread (210 denier) nylon netting in the 

 forward sections, with 6.4 mm stretched mesh 

 6-thread (210 denier) nylon netting at the cod 

 end (Higgins, 1970). This 3-fold difference in 

 mesh size corresponds with a nearly 3-fold dif- 

 ference in the size of skipjack which makes up 

 the two components of the catch (100% appar- 

 ent mesh retention at lengths greater than 

 15.5 mm and 37.5 mm, respectively, from curve 

 B of Figure 10). The Cobb trawl is, in eflfect, 

 two distinct nets which happen to be rigged in 

 tandem. The openings of the trawl and its cod 

 end during a tow are estimated to have radii of 

 about 4 m and 1.5 m, respectively, so that their 

 mouth areas differ by a factor of seven. The vol- 

 ume of water filtered by the cod end should thus 



814 



