WEBB: RESPONSES OF NORTHERN ANCHOVY LARVAE TO PREDATION 



change of the angle, «, subtended by the approach- 

 ing object measured at the prey's eye (Dill 1974a, 

 b). The looming effect is greater as the speed of an 

 approaching object increases, as the object gets 

 closer, and for larger objects. The threshold was 

 calculated at the start of an escape attempt by the 

 larvae, after Dill (1974a), 



do. 

 dl 



4 US 



h 



4(D + df +S^' 



(1) 



where 



U = predator speed at the time of 

 the prey response, 



Sh = predator shape; mean of maxi- 

 mum depth and width of the 

 clown fish, 



D = distance between the predator's 

 nose and the prey when the prey 

 responded, 



d = posterior distance of the preda- 

 tor's maximum depth and width 

 from its nose. 



Here, daldt at the start of the response by the 

 larvae is called the apparent looming threshold 

 (ALT). This is because the true threshold must 

 occur prior to the observed motor response because 

 there is a finite delay in the nervous system be- 

 tween receipt of the stimulus and the start of the 

 motor response. The value of this response latency, 

 or reaction time, is unknown for northern anchovy 

 larvae. 



cm larvae were not successful. Predator failures 

 are presumed due to predator error, which has 

 been reported as 89^^ for largemouth bass, Microp- 

 terus salmoides, (Nyberg 1971) and 34% for chain 

 pickerel, Esox niger (Rand and Lauder 1981). Lar- 

 vae that made an avoidance response to an attack 

 by clown fish were rarely pursued because clown 

 fish usually attacked another larva. Eight chases 

 were observed in 113 escape attempts by northern 

 anchovy larvae. One escape each was observed for 

 larvae in the 0.626 and 1.166 cm TL groups and two 

 in each of the 0.399, 0.867, and 0.953 cm groups. 



Larval responses to attacks by clown fish were 

 characteristic startle responses, followed by a 

 short period of sprint swimming. The startle re- 

 sponse consisted of a C-start form of a fast start 

 and a turn (Eaton et al. 1977). The sprint was a 

 period of swimming at constant speed for < 1 s (see 

 Hoar and Randall (1978) for definitions). Together, 

 these two components constitute a swimming 

 burst (Webb and Corolla 1981). 



The proportion of larvae responding to attacks 

 increased with size, from approximately 97c for 

 0.29 cm larvae to 80% for 1.2 cm larvae (Figure 

 lA). The proportion of larvae escaping an attack 

 similarly increased with size, but at a lower rate 

 than the proportion making an escape attempt 

 (Figure IB). This was because 26 ±10% {N = 10 

 groups of larvae) of larvae showing an escape re- 

 sponse were eaten anyway, before a chase began. 

 These larvae were defined as attempting to escape 

 too late. This proportion of escaping larvae caught 



RESULTS 



Clown fish fed discontinuously on eggs and lar- 

 vae, feeding intensely on prey in one location and 

 then swimming around the tank before feeding 

 again. Thus the clown fish did not attack prey as 

 encountered, and often ignored nearest prey be- 

 tween feeding bouts. The clown fish were continu- 

 ously active, swimming mainly by paddling 

 movements of the pectoral fins supplemented with 

 caudal fin beats. The mean speed was 3.2 ±0.5 

 cm/s {X±2 SE; N = 100; range 0.5 to 9.1 cm/s). 



Attacks were made on individual larva. Orien- 

 tation by the clown fish prior to a strike could not 

 be distinguished with confidence from normal 

 swimming. Prey that did not attempt to escape 

 were always caught, with the exception of eggs 

 and the smallest larvae tested. Thus 12% of strikes 

 on eggs and 3% of strikes on nonresponding 0.29 



08 



o 



O 



00, 



12 



T0T4L LENGTH (cm) 



FIGURE 1. — Relation-ships between (A) the proportion of north- 

 ern anchovy larvae responding to an attack by clovm fish and ( B) 

 the proportion escaping an attack, both as functions of the total 

 length. Circles are for prefeeding yolk-sac larvae. 



729 



