eggs to be about SS'/c/d. Thus, egg cannibalism 

 may be the cause of 32% of the natural egg mor- 

 tality in northern anchovy. 



Discussion 



We have not tried to trace the error terms 

 through the pyramid of calculations required to 

 estimate the proportion of natural egg mortality 

 attributable to cannibalism, although we give in 

 Table 2 the standard error of the mean where 

 estimates are available. The error in our estimate 

 most likely will be equivalent or higher than that 

 of the most variable parameter (i.e., mean eggs 

 per stomach and natural mortality of northern 

 anchovy eggsi. We have no estimate of the error 

 for the natural mortality of northern anchovy eggs 

 but mortality rates of pelagic fish eggs are known 

 to be high and variable; estimates range from 2 to 

 957c for various species (Jones and Hall 1974; 

 Vladimirov 1975). Regardless of the uncertainties, 

 we believe the results indicate that egg canni- 

 balism may be a major source of egg mortality in 

 the northern anchovy and a combination of patchi- 

 ness of eggs and selectivity in filter feeding may be 

 important in regulating the consumption of eggs. 



Cannibalism is a mechanism for density-de- 

 pendent regulation of fish populations (Gushing 

 1977), and egg cannibalism may be one of the 

 many mechanisms regulating the size of anchovy 

 populations. A simple model could be developed to 

 test this hypothesis if random filtering and a 

 random egg distribution were assumed. Although 

 the development of such a model is beyond the 

 scope of this paper, we wish to consider the extent 

 our observations differ from predictions based on 

 assumptions of randomness because this would be 

 a critical decision in the development of the model. 



The mean density of eggs in trawl associated 

 plankton tows was 32 eggs/m^ and the maximum 

 filtering rate of northern anchovy of mean weight 

 16.8 g, is 0.158 m^/h (Leong and O'Connell 1969). 

 For a northern anchovy (weight 16.8 g) to obtain 

 the estimated ration of 85.8 eggs/d, it would have 

 to filter continuously for 17 h at a density of 32 

 eggs/m^. Another approach is to estimate the 

 percentage of the volume of the habitat that could 

 be randomly filtered by a group of northern 

 anchovy schools. The mean weight of northern 

 anchovy schools in 20' grid squares is 2.05 x 10' 

 kg (excluding grid squares without northern an- 

 chovy) (Mais^); assuming the maximum depth of 

 eggs is 30 m (Hunter and Sanchez 1977), the 



volume of the habitat is 2.86 x 10^" m^ These 

 schools would have to filter for 24 h to consume 

 17% of the nightly egg production. The number of 

 hours of filtering required to obtain the average 

 daily ration of eggs in the first calculation or to 

 consume 17% of the egg production in the second is 

 too high. Thus a random encounter model does not 

 seem to account for the relatively high egg con- 

 sumption that was observed and patchiness and 

 selectivity in feeding may be the reasons. 



Northern anchovy eggs exist in patches (Hewitt 

 in press), probably in patterns similar to those 

 described for sardine eggs (Smith 1973). Our 

 observations of northern anchovy feeding on eggs 

 in the laboratory indicate that filtering may be 

 intensified when egg patches are encountered. 

 When we added a beaker containing anchovy eggs 

 to a 3.3 m diameter tank containing about 200 

 northern anchovy, they soon interrupted their 

 circuit of the tank, formed a tight mill at the site of 

 introduction, and filtered the region intensively. 

 If northern anchovy display such patterns of 

 behavior in the sea, the ration of eggs would be 

 expected to be higher than one predicted from 

 random filtering. Thus, present evidence indi- 

 cates that an assumption of randomness would be 

 unacceptable in a model to measure effects of 

 cannibalism as a regulatory mechanism. 



Larval cannibalism might also be significant. 

 Northern anchovy larvae are readily eaten by 

 adults in the laboratory. The absence of small 

 northern anchovy larvae in stomachs may have 

 been caused by the rapid rate of digestion (<30 

 min) and the low incidence of larger larvae (0.3% ) 

 caused by their low abundance in the sea. Thus, 

 cannibalism on larvae as well as that on eggs 

 might play a role in the regulation of northern 

 anchovy populations but additional information is 

 needed on feeding behavior, and on the size and 

 density of egg and larval patches, before an 

 evaluation of population effects can be made. 



Acknowledgments 



Sharon Hendrix and Harold Dorr (Southwest 

 Fisheries Center, National Marine Fisheries Ser- 

 vice, NOAA, La Jolla, Calif.) assisted in collection 

 and processing of adult northern anchovy samples. 

 Daylight trawl collections and associated plank- 

 ton tows were made by Anthony Koslow (Scripps 



''Mais, K. F. California Department of Fish and Game, 

 Cruise Reports 76-A-3 and 77-A-3. California Department of 

 Fish and Game, Marine Resources Region, Long Beach, Calif. 



815 



