708 



Fishery Bulletin 89(4), 1 99 1 



(15.1-17.8°C) flowed continuously through the tank, 

 except during the experiments when water flow was 

 stopped. Two household tungsten lamps above the tank 

 provided about 2000-3000 lux at the water surface. 



The experimental procedure was similar to Folkvord 

 and Hunter (1986). Experiments were run in the mor- 

 ning and early in the afternoon starting about 24 hours 

 after onset of starvation. Adult brine shrimp (Artemia 

 sp.) were used to identify variation in predator feeding 

 behavior between and during experiments. The addi- 

 tion of three brine shrimp or three starved anchovy 

 larvae or three fed anchovy larvae constituted a single 

 trial. Each experiment started with 3 Artemia trials 

 followed by 15 trials repeating the sequence; fed larvae, 

 starved larvae, and Artemia. The initial three brine 

 shrimp trials insured that the feeding performance of 

 the predators was stable, and subsequent brine shrimp 

 trials were to check for effects of satiation (Folkvord 

 and Hunter 1986). The brine shrimp trials were also 

 used to test for differences in feeding performance 

 among predator groups. No significant differences 

 existed among the 4 predator groups in the proportion 

 of brine shrimp eaten in 5 minutes (99% of the brine 

 shrimp were eaten), or in the proportion of fed or 

 starved larvae responding to attacks (logistic regres- 

 sions, P>0.40). 



A trial began when the prey were gently poured from 

 a beaker into the test container and ended when all 

 three prey were consumed or when 5 minutes had 

 elapsed. During a trial, each time a prey was attacked 

 by a predator, we recorded whether or not the prey 

 responded to the attack, whether or not it escaped, and 

 the time required to capture the prey. A predator at- 

 tack was defined as a change in swimming speed or 

 direction towards a prey followed by an opening of the 

 mouth. A response to a predator was defined as a 

 change in larval swimming speed or direction that 

 occurred during an attack. An escape was defined as 

 a predator attack in which the larva responded and 

 successfully avoided the predator. A total of 345 fed 

 and 345 starved larvae were used in the experiments, 

 but the results are based on the number of predatory 

 attacks and not on the total number of larvae tested. 

 Some larvae were attacked more than once and others 

 were not attacked at all. If a larva was attacked more 

 than once in an experiment, each interaction was con- 

 sidered a separate event. When a larva was taken 

 immediately after release or before it had moved, the 

 interaction was not assessed. The responsiveness to 

 attacks was assessed in 295 out of 325 attacks by 

 predators on fed larvae and in 266 out of 323 attacks 

 on starved larvae. 



Figure 1 



Proportion of fed and starved 9 mm SL northern anchovy 

 larvae that responded to attacks of adult northern anchovy 

 (Y), as a function of the duration of the starvation period (X 

 = elapsed time in days). Horizontal line represents average 

 proportion of fed larvae responding (0.4 ±0.03); logistic regres- 

 sion, effect of day, P>0.8. Dashed line for starved larvae is 



given by the equation: Y = e 



(-055-0.16-X) 



/(1 + e 



(-0.55-0.16-X) 



logistic regression, effect of day, P 0.044. Numbers next to 

 points are number of responses and interactions. 



Assessment of nutritional condition 



Histological criteria were developed to link starvation- 

 induced changes in avoidance behavior of larvae to 

 their nutritional condition. Samples of starved and fed 

 larvae from each of the three groups were taken for 

 histological analysis. Twenty starved and 10 fed larvae 

 were collected every morning starting about 24 hours 

 after onset of starvation. Larvae were fixed in Bouin's 

 solution, embedded in Paraplast-plus, serially section- 

 ed at 5^m, and stained with Harris hematoxylin eosin- 

 phloxin B. 



O'Connell (1976) found that pancreas condition, 

 notochord shrinkage and muscle fiber separation were 

 the histological characteristics that best distinguished 

 starving anchovy larvae from healthy ones during the 

 first feeding stage (4 mm SL, about 4 days posthatch). 

 These characteristics did not prove to be useful for the 

 diagnosis of starvation in the 9 mm (20 day posthatch) 

 larvae. For example, fibers of the trunk musculature 

 were perfectly arrayed in starved 9 mm larvae, and no 

 interfiber spaces existed in any of the larvae analyzed 

 even after 6 days of starvation. We also found no cor- 

 relation between the condition of the notochord and 

 starvation. Also presence, size, and abundance of hind- 

 gut vacuoles were interpreted as indicators of recent 

 feeding and not of larval condition. 



