WEBB: RESPONSES OF NORTHERN ANCHOVY LARVAE TO PREDATION 



sponse to attacks by largemouth bass and to 

 silhouettes. These data are not fully comparable 

 with those obtained here for several reasons. First, 

 the mean depth plus width was used to charac- 

 terize the shape of the clown fish. Dill ( 1974a) used 

 only width, citing reports that fish are more sensi- 

 tive to horizontal movement than to vertical 

 movement (Cronly-Dillon 1964; Jacobson and 

 Gaze 1964). In contrast, operant conditioning ex- 

 periments show that fish are particularly sensi- 

 tive to apices (e.g., Hinde 1970; Baerends 1971) 

 which occur at the dorsal and ventral margins of 

 laterally compressed bodies. In the absence of 

 definitive experiments relating shape to looming 

 response thresholds, the mean value of depth plus 

 width was considered most appropriate. Use of the 

 mean value of depth and width would give larger 

 values of da/dt than use of depth alone (Equation 

 (D). 



Second, the distance between the nose and the 

 maximum depth and width of the predator was 

 added to the reaction distance separating the pred- 

 ator and prey. This assumes that the prey either 

 has depth vision or sees the equivalent of a 

 silhouette of the predator. The inclusion of this 

 term would reduce values of da/dt compared with 

 Dill's method. 



The ALT values for postlarval zebra danios (2.0 

 cm long) was 0.43 rad/s. Northern anchovy larvae 

 of the same size would be expected to have a mean 

 value of about 0.6 rad/s from the relationship in 

 Figure 6. 



The overall function of the startle response is to 

 avoid predators. How effective is it? The clown fish 

 rarely pursued escaping larvae, although when 

 they did so, the larvae were easily caught. In these 

 experiments, chases may have been rare because 

 larval densities were high. However, observations 

 on adult piscivores attacking single prey show 

 that chases are also rare with pike, Esox ( Neill and 

 CuUen 1974; Webb and Skadsen 1980), largemouth 

 bass (Nyberg 1971; R W. Webb unpubl. obs.), and 

 rock bass, Ambloplites rupestris (P. W. Webb un- 

 publ. obs.). Presumably the cost of pursuit is large 

 relative to the benefits of capturing small prey 

 particularly where there are alternative prey. In a 

 normal planktonic assemblage, alternate prey 

 could be important in reducing vulnerability of 

 larval northern anchovy, especially in the pres- 

 ence of more opaque forms and those with more 

 strongly pigmented eyes (Zaret and Kerfoot 1975). 



Unfortunately, there are no field observations 

 on larval responses to predation, and the likeli- 



hood of making the requisite field observations is 

 remote. Nevertheless, the response of a prey to an 

 attack is an obvious indicator of the prey's aware- 

 ness and the possible difficulty of capture. The 

 reluctance of many predators to attack responding 

 prey, as noted above, together with the behavior of 

 the clown fish observed in these experiments, 

 imply that the startle response is an effective de- 

 terrent. Thus, it is most important that the larvae 

 respond, but initially maximum swimming speeds 

 are not required. Indeed the latter would be 

 energetically more costly. Larvae clearly behave 

 appropriately with a submaximal evasion (Figure 

 4), except when maximum performance becomes 

 desirable in the rare event of a chase. Neverthe- 

 less, timing of the escape attempt must be accu- 

 rate as 24 to 30^f of the larvae attempted a re- 

 sponse too late to escape capture. 



Larval looming response thresholds will not 

 only be important in escaping biting predators, 

 but also other predation threats. Webb and Corolla 

 (1981) discussed relationships between burst 

 swimming performance of northern anchovy lar- 

 vae and escape probabilities from plankton nets as 

 a crude analogy with filter-feeding predators. 

 While swimming performance could explain a 

 large part of net avoidance, other factors were 

 involved. Webb and Corolla suggested that declin- 

 ing response thresholds with experience would be 

 important so that larger larvae responded earlier to 

 an impending collision. The inverse relation be- 

 tween ALT with larval total length suggests that 

 such changes occur. Presumably, similar 

 thresholds or size relations would apply to larger 

 predators. Then the reaction distance to a net mov- 

 ing, for example, at a given towing speed would be 

 expected to be greater for larger northern anchovy 

 larvae. This would contribute significantly to the 

 size-dependent sampling bias of such nets. 



This work has attempted to evaluate a method 

 for quantifying responses of a fish larva to attacks 

 by a predator as one step in studying the neglected 

 aspect of predation on larval mortality. The ad- 

 vantages of the method are the visualization of 

 both opaque and transparent individuals of small 

 size and continuously recording their behavior. 

 The disadvantages are that the space viewed must 

 be small and hence only small predators can be 

 used, and filtering predators are excluded. How- 

 ever, predation could be studied for particular 

 feeders (e.g., biting fish and chaetognaths) ar- 

 thropods (e.g., copepods and euphausids), and less 

 discriminating feeders such as thaliaceans and 



733 



