FISHERY BULLETIN: VOL. 79, NO. 1 



cruise (August 1975), the schoors feeding on both 

 the small copepods, A. tonsa, and the chaetognath, 

 Sagitta spp., increased as a single function of prey 

 weight (Figure ID). However, the school appeared 

 to select the larger copepod, Calanus pacificus, 

 over the other prey examined from this cruise. To 

 test for the differential feeding selectivity for C. 

 pacificus, a single linear regression was per- 

 formed through the pooled data for prey consump- 

 tion from the sampling period (Figure ID) (Quade 

 1967). All data points for C. pacificus and only 3 of 

 14 data points for the consumption of Sagitta spp. 

 and A. tonsa lie above this regression line, indicat- 

 ing a significantly heterogeneous distribution of 

 the data (X2 = 11.57; P< 0.01). 



For the two sets of samples collected from two 

 different schools on consecutive days during the 

 March 1976 cruise, no significant differences were 

 found between the control tows and those taken in 

 the wake of the school either in the displacement 

 volumes or in the plankton's size-frequency com- 

 position. This cruise was undertaken during an 

 intense spring diatom bloom; ambient plankton 

 concentrations were a factor of 20 greater than 

 during any other cruise. At these plankton densi- 

 ties, northern anchovies could fill their stomachs 

 [ca. 5% of body weight (Rojas de Mendiola and 

 Ochoa 1973) for Engraulis ringens or 0.05 g C] in 

 approximately 40 min by simple filtration (fil- 

 tering rate per individual northern anchovy = 

 2 1/min (Leong and O'Connell 1969)). Thus, the 

 schools, or some part of them, may have ceased 

 feeding. Furthermore, at these high densities of 

 plankton, the schools would have to ingest far 

 more material than on the previous cruises to 

 consume a detectable fraction of the plankton. The 

 fish stomachs examined from both schools were 

 full, but the data on the dimensions and biomass of 

 the schools indicate they were not significantly 

 more densely packed. The schools had thus appar- 

 ently been feeding, at least intermittently, but 

 under these conditions, their feeding selectivity 

 could not be determined from the plankton sam- 

 ples alone. 



To analyze the northern anchovy's feeding dur- 

 ing this cruise, I compared the size-frequency 

 composition of prey in the stomach contents with 

 that found in the zooplankton tows (Table 3). The 

 data were analyzed using Ivlev's Electivity Index: 

 E = (r - p)l{r +p), where r = the proportion the 

 prey item represents in the diet and p = the 

 proportion the prey represents in the plankton 

 samples (Ivlev 1961). 



138 



The feeding of the northern anchovy on the 

 cladocerans (predominantly Evadne nordmanni), 

 which dominated the plankton during the spring 

 bloom was a linear function of prey size (Figure 

 IE). No significant difference was found between 

 the electivity of the two schools sampled under 

 similar conditions of food density and composition 

 on this cruise, although northern anchovies from 

 the school of 8 March 1976 were the largest and 

 those sampled on the following day were the 

 smallest encountered during the study. Their 

 difference in mean length {x = 122.84 and 98.25, 

 respectively) indicates the schools were composed 

 predominantly of Il-group and 0-group fish, 

 respectively, with a difference in mean weight of 

 approximately a factor of 2 (calculated from Saka- 

 gawa and Kimura 1976). But this difference in 

 size apparently did not lead to a significant 

 difference in their feeding selectivity under the 

 sampling conditions. 



Comparison of 

 Feeding Selectivity Between Cruises 



The three northern anchovy schools studied on 

 the cruises of August 1975 and April and October 

 1976 each consumed approximately 100% of the 

 largest prey available and a small fraction of the 

 smaller prey. However, because the size distribu- 

 tion of available prey varied greatly between 

 cruises, prey items that were almost entirely 

 removed from the water when only small prey 

 were available (e.g., the later copepodite stages of 

 small copepods, such as A. tonsa or P. parvus, 

 encountered during August 1975 or October 1976) 

 were virtually ignored when larger prey were 

 present (e.g., on the cruise of April 1976, compare 

 Figure lA, C with Figure ID). 



The prey size at which the northern anchovy 

 school's consumption was approximately 100% on 

 these three cruises (which may be defined as the 

 intersection of the linear regressions with the line 

 y = 1) varied by more than a factor of 100 (Table 5; 

 Figure lA, C, D). Furthermore, while the school's 

 feeding selectivity was consistently a positive 

 function of the prey's size, the slopes of the linear 

 regressions from the cruises of August 1975 and 

 April and October 1976 also varied by more than 

 two orders of magnitude (Table 5). Both factors 

 are related to the size range of prey available to 

 the anchovy on these cruises, which varied to a 

 similar degree. 



There is a positive relation between the prey 



