FISHERY BULLETIN: VOL. 79, NO. 1 



Table 2. — Characteristics of northern anchovy schools examined for feeding selectivity: overall school biomass, physical dimensions, 

 velocity and prey consumption, number measured, and mean and range of standard lengths oi EngrauUs mordax within the schools. 



'Percentage body-weight consumed/h = [(volume swept clear/h)(food removed/unit volume)/(school biomass)] x 100 



= [(7r(0.5b)(0.5d)(e x m^)(f x 10-3)/(a x 10« x 0.05)] x 100. 



^Estimated from physical dimensions, density of 1.5 kg/m^ (Hewitt etal. 1976), and assuming actual shape to be cylinder (August 1975) or oblate spheroid (April 

 1976). 



^Fisherman's estimate based on purse seining of school. 



"Back-calculated from tonnage, physical dimensions, and assumptions of average school density = 1.5 kg/m^ and of a cylindrical (October 1976) or oblate 

 spheroidal (March 1976) school shape. 



^Estimate. 



evaluate relations between feeding conditions 

 and school size and configuration. 



Samples of the northern anchovy were obtained 

 from four of the schools. Two of the schools were 

 composed of northern anchovies that had com- 

 pleted approximately 1 year's growth (98-99 mm 

 SL); the other schools were composed of predomi- 

 nantly I-group (115 mm SL) and Il-group (123 mm 

 SL) northern anchovies (Table 2; Sakagawa and 

 Kimura 1976). Since the schools composed of the 

 largest and smallest fish were sampled under the 

 same feeding conditions (i.e., the spring diatom 

 bloom dominated by cladocerans of March 1976; 

 Tables 1, 3), the feeding selectivity of large and 

 small northern anchovies can be compared. 



Table 3. — Ivlev's Electivity Index (E) as computed from the 

 frequency of size classes of Evadne spp. examined in northern 

 anchovy stomach contents and plankton tows taken in vicinity of 

 the two schools sampled on 8 and 9 March 1976. 



Feeding of Northern Anchovy Schools 



The impact of the schools' feeding could be 

 clearly determined from the plankton samples 

 during all sampling periods except those of March 



1976. In all 40 size categories of prey enumerated 

 from August 1975 and April and October 1976 

 cruises, the median concentration was less in tows 

 taken in the wake of the school than in control 

 tows (Table 4). These data were analyzed as the 

 fraction consumed [= 1 - (density of organisms in 

 wake of school)/(density in controls)] as a function 

 of the prey organisms' body size. 



In computing regressions to analyze the feeding 

 selectivity of the northern anchovy schools, it was 

 often not clear either by eye, through analysis of 

 residuals, or from the significance level of the 

 regression whether a linear or curvilinear rela- 

 tionship best fit the data (Figure 1). In these 

 instances, two regressions were performed: a lin- 

 ear regression and a regression in which the 

 independent variable (i.e., prey body size) was 

 loge-transformed. In computing the linear regres- 

 sions (Figure 1), data points are excluded past the 

 first size class at which the school has effectively 

 consumed all the plankton (i.e., when consump- 

 tion is >90%). An arcsine transformation was not 

 performed, although it has been recommended for 

 regressions performed on data expressed as frac- 

 tions or percentages (Sokal and Rohlf 1969). The 

 arcsine transformation did not significantly affect 

 the form of the regressions presented below and 

 only slightly enhanced their significance level. 

 The data are therefore presented untransformed 

 (Figure 1). 



There was consistently a significant positive 

 relationship between the fraction consumed and 

 the size of the anchovy's prey on the cruises 

 (Figure 1), despite the diversity of prey items 

 within each cruise and the considerable differ- 

 ences in the composition and density of the zoo- 

 plankton between cruises (Tables 1, 4). The frac- 

 tion consumed ranged from 10-30% for the 

 smallest organisms enumerated to 95-100% for 



134 



