FISHERY BULLETIN: VOL. 70, NO. 3 



CATCH/ 300 m' 



1 



Figure 4. — Analysis of length-frequency data for day- 

 light catches of Stolephorus from 44 paired samples. 

 Purse seine data (triangles), a measure of absolute 

 abundance, are approximated by the straight line, Nl- 

 Catch with the 1-m net, C^, has been corrected for 

 changes in population (N) with length (L), to obtain 

 the middle curve, Ci/N^, which shows the minimum 

 probability of capture, P^ as a function of length for the 

 1-m net. Three theoretical curves for P^ at constant 

 relative reaction distance (xq/R) are reproduced from 

 Figure 3, to show degree of fit between data and theory. 

 Data from Murphy and Clutter (1972). 



converted data can now be directly compared 

 with the theoretical curves of Figure 3, three of 

 which are shown on Figure 4 (the curves for Pc 

 at Xo/R = 5, 6, and 8). 



The adjusted 1-m net catch data fall near the 

 theoretical curves for Pc at relative reaction 



distances, Xo/R, of 6 and 8; since i? = 0.5 m 

 for this net, the reaction distance of Stolephorus 

 apparently ranges between 3 and 4 m. Murphy 

 and Clutter (1972) obtain comparable results 

 from an equation they derive for calculating Xo 

 directly. They used similar assumptions — in 

 particular, an escape speed of lOL per second — 

 but set Pc = 1.0 for the third class interval 

 (3.5 mm). Here Pc was set at 1.0 at the inter- 

 cept of the purse-seine catch curve with the or- 

 dinate axis, i.e., Pc = 1.0 at No, the fictitious pop- 

 ulation density at zero length, because only at 

 this length does the assumed relationship be- 

 tween speed and length yield zero velocity. Since 

 Nl, the population density at any given length L, 

 is equal to No for the adjusted data, the above 

 procedure is internally consistent. 



There are two exceptions to the otherwise 

 fairly good fit between the adjusted data and 

 the theoretical curves. The first, catch with Pc 

 exceeding 1.0 for the third class interval (3.5 

 mm), may be due to sampling variability, over- 

 estimates of the water sampled by the purse 

 seine, or underestimates of the water sampled 

 by the 1-m net. It could also be due to less than 

 optimum avoidance behavior by these small fish. 

 The second exception, unexpectedly large catch- 

 es in the three largest class intervals (12.5-14.5 

 mm) could be due to variability, or to a decrease 

 in mean reaction distance for these animals, 

 from a "normal" value of 3-4 m down to about 

 2.5 m {xo/R — 5). Isaacs (1965) suggested 

 that daytime tows should catch more than pro- 

 portionate numbers of the sick, lame, or lazy; 

 this effect should be most pronounced for the 

 largest animals, which normally are the ones 

 best able to dodge the net. 



Minor fluctuations in the fit of the adjusted 

 1-m net data on Figure 4 are probably due to 

 sampling variability, but to illustrate use of 

 avoidance theory, suppose that the minor peaks 

 and valleys for fish of intermediate sizes were 

 significant. Note that these points tend to fall 

 into two groups: those near the curve for Pc 

 at Xo/R = 6, and others, Xo/R = 8. This could 

 be an artifact due to schooling behavior, since 

 group reaction should reduce the probability of 

 capture below the theoretical minimum value for 

 individual reaction. One measure of this effect 



806 



