FISHERY BULLETIN: VOL. 70, NO. 3 



(Nl) by a factor of 10 or more, as it does in 

 Figure 4. 



When this assumption is made, however, it 

 must be recognized that the results of any sub- 

 sequent analysis are contaminated by whatever 

 population structure effects are present and 

 erroneously attributed to avoidance. 



Anchovy (Engraulis mordax) larvae were 

 also collected off southern California by Murphy 

 and Clutter (1972) with a 1-m net and an Isaacs- 

 Kidd midwater trawl (IKMT). Both nets had 

 similar mesh size and were towed in daylight 

 and at night at comparable speeds, nominally 

 75 cm /sec (although the IKMT may have been 

 towed somewhat faster) . Figure 5 shows data 

 from these tows, converted to speed frequency 

 by assuming that Ue = lOL. No estimates of 

 absolute abundance were made for anchovy by 

 Murphy and Clutter; the plankton purse seine 

 was not used in these trials. However, Figure 4 

 shows that even 10-fold changes in Nl, abund- 

 ance as a function of length, do not materially 

 alter the shape of the Stolephoriis speed-fre- 

 quency curve. Essentially identical results 

 could have been obtained by fitting the catch data 

 to the theoretical curves without correction for 

 changes in population density. Since anchovy 

 are in many ways similar to Stolephorus, it may 

 be tentatively assumed that avoidance is the 

 major determinant of size-frequency curves for 

 anchovy, as it clearly was for Stolephorus (Fig- 

 ure 4). 



Without an independent estimate of absolute 

 abundance, there is no priori relationship be- 

 tween catch and Pc. This relationship must 

 therefore be established by seeking the best pos- 

 sible fit between theory and observations. Speed- 

 frequency values were plotted on tracing paper, 

 using the semilog coordinate system of Figure 3. 

 The resulting graph was placed over the upper 

 panel of Figure 3 and moved up and down until 

 a good fit resulted (right-left motion is not al- 

 lowed, since the position of points along this axis 

 is determined by size, the size-speed relation- 

 ship, and net speed). This procedure is equiv- 

 alent to making various assumptions about A^o, 

 the fictitious population density at zero length, 

 for which Pc = 1 by definition since Ue = 0. 

 When this is done, the speed-frequency curves 



become speed - Pc curves, showing probability 

 of capture for each class interval. Best fits were 

 obtained for the 1-m net anchovy catches with 

 No = 75 anchovy per 100 ml wet plankton (left 

 panel. Figure 5). For the IKMT, the best fit 

 yielded No — 60 anchovy per 100 ml wet plank- 

 ton (right panel, Figure 5) . It might have been 

 worthwhile to treat day and night catches sep- 

 arately, as was done for the two types of gear, 

 but the resulting diflferences in A^o between day 

 and night tows are small, change sign for the 

 two different nets, and probably are not sig- 

 nificant. 



Figure 5 shows that the variance for anchovy 

 data was smaller than that for Stolephorus 

 catches (Figure 4), so that the anchovy data 

 fit the theoretical curves somewhat better. There 

 is no way to tell whether this difference in var- 

 iance is due to differences in sampling or popu- 

 lation structure. For both nets, day-night dif- 

 ferences in xo/R are small, but reaction distances 

 tend to be slightly greater for tows taken during 

 the day, as might be expected. A surprising re- 

 sult is the similarity in values of Xo/R obtained 

 for the two nets: about 6.6 for the 1-m net and 

 5.4 for the IKMT (if slightly lower values for 

 the largest size classes are ignored) . This yields 

 reaction distances, Xo, of 3.3 m for the 1-m net 

 and 8.2 m for the IKMT. Clearly anchovy re- 

 acted to the IKMT at considerably greater dist- 

 ances. A light dashed line on Figure 5 (right 

 panel) shows the dramatically increased catches 

 which would be expected if anchovy had reacted 

 to the IKMT at 3.3 m (xo/R ~ 2), as they ap- 

 parently did in the case of the 1-m net. 



The analysis shows that maximum values of 

 Pc for the IKMT amounted to only 0.25, as com- 

 pared to maximum values of Pc for the 1-m net 

 of 0.40 or more. 



Mesh retention appeared to diflfer for the two 

 nets, despite similarity in mesh size (according 

 to Murphy and Clutter, 1972, the IKMT was 

 meshed with Nitex," and the 1-m net with gauze 

 silk, 56xxx grit, having openings of 0.505 and 

 0.55 mm, respectively). Retention dropped be- 

 low lOO'^r, based on deviations from the the- 



" Reference to commercial products does not imply 

 endorsement by the National Marine Fisheries Service. 



808 



