FISHERY BULLETIN: VOL. 87. NO. 3, 1989 



Similarity of growth and starvation indices at 

 the tvi^o sites indicates that most larvae had en- 

 countered sufficient forage to sustain "normal" 

 growth rates, or that the fraction that had not 

 done so were quickly removed from the system, 

 i.e., not represented in our collections. Food 

 abundance estimated from integrative tows may 

 not reflect the actual availabihty of food to larvae 

 because microplankton prey frequently occur in 

 patches and laminae of very small extent (cf. 

 Owen 1989). Also, larvae may have been sus- 

 tained on food that passed through the net. 

 Fewer larvae were collected at Site 2 than at 

 Site 1, but starvation incidence was at least as 

 low there as at Site 1. Furthermore, Site 2 

 larvae were growing as fast as those at Site 1, 

 even though average microplankton concentra- 

 tion at Site 2 was less than half that at Site 1. 

 Despite lower average food concentrations, 

 there were zones at Site 2 that evidently con- 

 tained enough food to support growth of the 

 larvae. 



An analogous result is given by Butler (1989), 

 which showed that periods of diminished forage 

 ■production, such as el Niiio, have no discern- 

 ible effect on growth rates of field-caught lar- 

 vae. But Theilacker and Watanabe (1989) 

 showed by experiment that starvation measur- 

 ably retards larval growth. This apparent para- 

 dox is resolved if, in the sea, larvae that survive 

 to be sampled have found enough food to gi-ow 

 at normal rates even in abnormal periods, 

 whereas larvae that have been deprived to the 

 extent that their growth is adversely affected 

 soon vanish, perhaps by predation owing to 

 their weakened condition rather than by starva- 

 tion directly. 



In contrast with the common view that north- 

 ern anchovies broadcast sex products indis- 

 criminately with the strategy that some get 

 lucky, we advance the hypotheses that an- 

 chovy spawn where their offspring are equally 

 hkely to survive, even under widely different 

 environmental conditions, and that these con- 

 ditions mediate their spawning intensity. This is 

 the central theme of MacCall's (1983) "Basin 

 Model" of habitat selection by the northern 

 anchovy. 



The Basin Model postulates that when a popu- 

 lation is large, its adults occupy less suitable 

 habitats in which mortality of spawn per capita 

 tends to equal that in more favorable habitats. 

 This occurs because cannibalism of spawn by 

 adults is higher in favored habitats than in pe- 

 ripheral habitats. 



We roughly partition sources of overall larval 

 mortality ((3/) into that due to cannibalism ((3,.) 

 and that due to other causes (other predation, 

 3p, and starvation, p,,). Diffusive change, a 

 source of "apparent" mortality, is assumed from 

 physical arguments above to be the same at both 

 sites. Thus, 



P, = (3, + Pp + p, . 



If higher egg concentration denotes higher 

 adult occupation, p,. was greater at Site 1 than at 

 Site 2. Anchovy egg concentration at Site 1 aver- 

 aged 37/m■^ 15 times the egg concentration at 

 Site 2. This difference is too great to be attrib- 

 uted to differences in batch fecundity of the 

 spawners, which varies by a factor of about two 

 (Hunter et al. 1985). 



For larval mortality, P/, to have been equiv- 

 alent at the two sites, mortality from other 

 sources (Pp + p.,) is required to have been 

 greater at Site 2 than at Site 1 to the degree of 

 offsetting the difference between sites in par- 

 ental consumption of spawn (p^.). But neither 

 larval growth rates nor starvation incidence dif- 

 fered between sites, showing that their p.,. values 

 were about equal. This being so, cannibalism 

 was offset by other predation rather than by 

 starvation. This requires increased predation 

 (Pp) at Site 2 over that at Site 1. 



For comparison of p,, between sites, we 

 formed rough indices of predation pressure, P, 

 from catches by vertical net tow pairs. P is the 

 concentration of the five most numerous preda- 

 tors caught by the 333 ixm mesh net, divided by 

 concentration of anchovy eggs and larvae caught 

 in the corresponding 150 \x.m mesh net. Predator 

 populations in the samples consisted mainly of 

 raptorial copepods and chaetognaths. Mean 

 values of P were 2.8 predators/anchovy at Site 1 

 (26 tows) and 5.5 predators/anchovy at Site 2 (24 

 tows). This difference indicates compensatory 

 predation at Site 2. Confirmation of compensa- 

 tory predation, however, is not possible from 

 this data set because our nets missed larger, 

 faster predators such as euphausiids, and be- 

 cause species-specific and size-specific predation 

 rates are largely unknown. 



CONCLUSIONS 



Among the several characteristics of the 

 ichthyoplankton investigated at contrasting 

 habitat sites, only one, larval production rate, 

 was clearly different between sites. Growth and 



686 



