FISHERY BULLETIN: VOL. 79, NO. 1 



the school from a considerable distance, which 

 seems unlikely, they could not escape it. 



There are also advantages to the direct field 

 sampling method: 1) It circumvents uncertainties 

 attendant upon extrapolating from laboratory to 

 field conditions. 2) It avoids the bias inherent in 

 the use of fish stomachs, due to the differential 

 digestibility of various prey organisms. This can 

 significantly bias estimation of selective predation 

 (e.g., see Gannon 1976). 3) It permits study of far 

 larger aggregations than can be accommodated in 

 the laboratory In the Southern California Bight, 

 approximately 90% of the biomass of the northern 

 anchovy population is found in schools >25 t 

 (calculated from Hewitt et al. 1976) — the size of 

 the schools sampled in this study. 



More generally, the feeding of planktivorous 

 fish has hitherto been studied experimentally 

 using individual or small groups of fish (Ivlev 

 1961; Beukema 1968; Leong and O'Connell 1969; 

 O'Connell 1972; Ware 1972; Werner and Hall 

 1974; Confer and Blades 1975; O'Brien et al. 

 1976). Prey selectivity was analyzed as a complex 

 function requiring understanding of the preda- 

 tor's metabolic state and the prey's size, density, 

 and patchiness. It is an awesome, if not impossi- 

 ble, task to extrapolate this analysis to field 

 situations involving 10^-10' predators interact- 

 ing within a school as they feed upon natural prey 

 assemblages. 



A field-oriented approach permits study of the 

 feeding of fish schools from the perspective of a 

 higher level of organization, as well as permitting 

 a test of laboratory studies. Vievdng the method in 

 this context, it is important to note the close 

 replication of results in the instance in which it 

 was successfully repeated under a similar set of 

 field conditions (March 1976, Figure IE). The 

 median significance level of the regressions (and 

 the regression coefficients) in Figure 1 was also 

 quite high (P< 0.025), despite the relatively few 

 data points on which the regressions are based and 

 the nature of the data, which is derived from field 

 sampling of zooplankton. The feeding of fish 

 within large schools is sufficiently consistent that 

 significant proportions of the variance in plankton 

 distributions in the wake of the schools can be 

 explained in terms of their impact. 



Comparison with Other Studies 



The results of the present study appear consis- 

 tent in broad outline with the pattern of size- 



selective feeding noted in laboratory studies; i.e., 

 prey selectivity proved a function of prey size 

 rather than of their taxonomy (Ware 1972; 

 O'Brien et al. 1976). This was found for a range of 

 prey sizes and taxa (Figure 1, Table 4). Signifi- 

 cantly enhanced consumption of a particular 

 species was observed only once; i.e., for Calanus 

 during the April 1976 cruise (Figure ID). Since 

 Calanus was encountered only on this cruise, it is 

 not clear how much significance should be 

 attached to this finding. 



This study indicates it may be possible to predict 

 the feeding selectivity of northern anchovy schools 

 on the basis of data on the size of prey available to 

 the school in significant quantity; i.e., both the 

 slope of the northern anchovy's feeding selectivity 

 and the size of the prey that will effectively be 

 removed entirely from the water is a function of 

 the largest available prey (Figure 2). This result is 

 attractive, since data on the size distribution of 

 the zooplankton can now be collected on a routine, 

 continuous basis (Mackas and Boyd 1979; Herman 

 and Dauphinee 1980). 



However, experimental studies have often 

 found that the concentrations of individual prey 

 items significantly influence the feeding selec- 

 tivity of fish (Ivlev 1961; Beukema 1968; O'Con- 

 nell 1972; Ware 1972). This is presumably medi- 

 ated through the varying degrees of experience the 

 predator will have with prey at different densities. 

 However, prey density did not appear to be a 

 significant factor in determining the northern 

 anchovy's prey selectivity in this study. On both 

 the August 1975 and April 1976 cruises, small 

 copepods and chaetognaths of equivalent body 

 weight were consumed in approximately equal 

 proportions (Figure lA, D), although their rela- 

 tive densities frequently varied by a factor of 5 or 

 greater (Table 4). Nor can prey density be invoked 

 to explain the anchovy's heightened selectivity for 

 Calanus (Figure ID). The density of Calanus 

 (April 1976) did not appear to differ significantly 

 from the density of Acartia and Sagitta of sim- 

 ilar size that were consumed to a lesser degree 

 (Table 4). 



This apparent difference between experimental 

 and field results may arise from significant differ- 

 ences in the distribution of prey typically avail- 

 able in the two situations. In laboratory studies 

 examining the influence of prey density on feeding 

 selectivity, the fish are typically offered several 

 prey items at varying densities (Ivlev 1961; Beu- 

 kema 1968; O'Connell 1972; Ware 1972). Under 



140 



