SMITH ET AL.; MORTALITY RATES OF SARDINE AND ANCHOVY EGGS OFF PERU 



in the initial period was higher owing to an inter- 

 action between local density and mortality rate 

 on eggs at the highest distributional densities 

 (McGurk 1987). 



The Central Limit Theorem requires indepen- 

 dence among sample values. With the spawning 

 of Pacific sardine and Peruvian anchovy off Peru 

 compressed against the coast in a 30 mile band in 

 1981, abundance estimates from stations sepa- 

 rated in the cross-shelf dimension at 3 mile inter- 

 vals are not likely to be independent. The effect 

 will be an underestimate of sample standard 

 error of the mean (Table 2). The degi-ee of the 

 same type of error in the along-shelf plane was 

 probably smaller because the transects are 10 

 nmi apart. Also the collection of sites used for 

 the estimates of abundance of 1 and 2 d old eggs 

 in the first half of each day (Al and Bl in Table 2) 

 are from different stations from those used for 

 estimates of abundance of 1 and 2 d old eggs in 

 the second half of each dav (A2 and B2 in Table 

 2). 



Other Mechanisms 



Egg cannibalism is probably only one of sev- 

 eral possible population controlling mechanisms 

 in anchovies. For example, cannibalism could 

 occur at life stages other than the egg stage. 

 There have been recent observations, in the lab- 

 oratory, of older anchovy larvae consuming 

 younger larvae (Brownell 1985). In these obser- 

 vations, the larvae became more vulnerable as 

 pigmentation formed in their eyes and integu- 

 ment. 



Energy demands for egg production and 

 spawning could lead to aggi-egations of adults 

 where food production and, coincidentally, egg 

 mortality are both high. That, in both Pacific 

 sardine and Peruvian anchovy, higher rates of 

 egg mortality and egg production occun-ed in the 

 same region (IMARll) seems to support this 

 view. Lastly, from laboratory observations, 

 northern anchovy larvae, only a day old, may 

 seek and maintain position in patches of Gym- 

 nodinium, making these larvae vulnerable to 

 enhanced feeding activity by many other organ- 

 isms in the same patch or layer (Hunter and 

 Thomas 1974). 



size of the recruited stock is not determined in 

 the first 20 days following spawning. On the 

 other hand, Pauly"' concluded that egg mortahty 

 due to cannibalism by the adult stock is a pri- 

 mary density dependent control on recruitment 

 in the Peruvian anchovy, Engraulis ringens. 

 Smith (1985) lists three situations in which can- 

 nibahsm by schooling pelagic spawners would be 

 an effective population controlling activity: 1) 

 directed filtering behavior on dense aggi'ega- 

 tions of eggs; 2) encounter with other schools' 

 patches of eggs; and, 3) anchovy population 

 switching from biting to filtering behavior. 



Directed Filtering 



In the first few hours after spawning, before 

 much dispersal has taken place, directed filter- 

 ing could be an important source of mortality 

 (Smith and Hewitt 1985b; McGurk 1987). Hunter 

 and Dorr (1982) found that adult northern an- 

 chovy filtering was induced by 5 northern an- 

 chovy eggs/L and sustained by 1 or 2 eggs/L in 

 laboratory tanks. Given that the level of artificial 

 feeding is higher for laboratory animals and that 

 probably the quality of water for sensing prey is 

 lower, these thresholds may be higher in the 

 laboratory than in the sea. One disadvantage of 

 this mechanism for population size control is 

 that, because of proximity, the most likely school 

 to encounter the newly spawned patch is the 

 school from which the spawn was produced. 

 Santander et al. (1983) found that Peruvian an- 

 chovy eggs < 2 hours old were overrepresented 

 by a factor of three in the stomachs of Peruvian 

 anchovy relative to their incidence in the sea. In 

 a typical spawning school (Santander et al. 1984; 

 Alheit 1985), there would be 10 times as many 

 Peruvian anchovies feeding on eggs than females 

 spawning eggs. While the mortality rate might 

 be sufficient, it is difficult to seen how a change, 

 in population size or in density over thousands of 

 square kilometers, could materially affect a 

 species grazing for a few hours on its own eggs 

 over a range of a few hundred meters. This 

 source of mortality, although large, may not be 

 sufficiently variable with population density to 

 control recruitment. 



Egg Cannibalism and Population 

 Control 



For the northern anchovy, Engraulis mor- 

 dax, Peterman et al. (1988) concluded that the 



'The estimate of instananeous mortality "z" derived by 

 Pauly (1987) is strongly biased downward at low mortality 

 rates but should still be valid at high levels of parent stock. 

 Daniel Pauly, International Center for Living Aquatic 

 Resources Management, Manila, PhiHppines, pers. commun. 

 March 1989. 



503 



