FISHERY BULLETIN: VOL. 79, NO. 1 



euphausids has been documented (Lillelund and 

 Lasker 1971; Theilacker and Lasker 1974), but 

 since experiments were performed in the dark, the 

 factors leading to reduced vulnerability are not 

 known. An alternative approach to assessing the 

 importance of changing swimming performance 

 with larval size is to consider an artificial "pred- 

 ator," a net, which simulates some actions of cer- 

 tain natural predators, such as filter feeders. The 

 problem of sampling bias by nets has provided a 

 variety of observations and models suitable to 

 evaluate size-performance effects on vulnerabil- 

 ity. 



Most studies of net avoidance start with, or par- 

 allel, Barkley's (1964) model on selectivity of 

 towed nets. This model identified three factors de- 

 termining escape (and catch) probabilities: reac- 

 tion distance, speed, and orientation of the escape 

 path. Endurance must be added. In the absence of 

 adequate data on the relevant parameters for fish 

 larvae, results of sensitivity analyses have been 

 fitted to catch data (Barkley 1964, 1972). The pres- 

 ent data can be used to compare scaling relation- 

 ships between escape probability and burst 

 swimming performance to evaluate the impor- 

 tance of the latter in determining vulnerability of 

 northern anchovy larvae to net "predators." Such a 

 scaling approach allows some simplification be- 

 cause it is reasonable to assume that probability 

 distributions for orientation of escape paths do not 

 scale with size. 



Escape probabilities, as a function of length, can 

 be estimated from a comparison between day and 

 night catches using towed nets. The ratio of 

 night/day catches as a function of larval northern 

 anchovy length is well known to be linear (Ahl- 

 strom 1954; Zweifel and Smith in press) which 

 means that the relationship between escape prob- 

 ability and size is curved, rapidly approaching an 

 asymptote (Figure 6). The data shown in Figure 6 

 were obtained with aim diameter ring net towed 

 obliquely at about 125 cm/s (Zweifel and Smith in 

 press). Ninety percent avoidance is reached by 1.0 

 cm larvae, with rather small improvements in 

 escape probability at greater lengths. The decline 

 in escape probability cannot be attributed to the 

 capture of starving or sick larvae (Isaacs 1965) 

 because O'Connell (1980) found no emaciated 

 larvae >1 cm TL in net-caught samples. 



In order to compare performance data with the 

 catch probabilities, mean speeds and distances 

 traveled in a burst were normalized about the per- 

 formance measured for 1 cm larvae (Figure 6). The 



06 0,8 10 12 



TOTAL LENGTH (cm) 



1,6 



1.8 



FIGURE 6. — Escape probabilities, estimated from day-night 

 plankton net hauls, and those possible on the basis of burst 

 swimming speed, burst endurance, and their product as a func- 

 tion of total length of northern anchovy larvae. Performance 

 data are normalized with respect to a larval length of 1 cm, where 

 the escape probability reaches 90%. Data for day-night catches 

 were taken from Zweifel and Smith (in press). 



scaling relationship for mean speeds in a burst, 

 assuming bursts are repeated until a larva es- 

 capes, obviously do not parallel escape prob- 

 abilities (Figure 6). Endurance in a burst shows a 

 greater length effect, which is further accentuated 

 by the interaction (product) of speed and endur- 

 ance. This latter curve has the steepest slope, most 

 closely paralleling escape probability, such that it 

 is apparent that speed and endurance are major 

 determinants of avoidance ability. 



However, the shape of the curves suggests avoid- 

 ance is underestimated for larger larvae between 

 about 0.4-1.0 cm long. Length dependent matura- 

 tion of sensory systems and motor control (O'Con- 

 nell 1980) may contribute to increased escape 

 probability of larger larvae. A behavioral factor 

 may also contribute. Dill (1974) has shown that 

 experienced juvenile zebra danio respond earlier 

 to a predator than do inexperienced fish. Since the 

 predation rate is high for northern anchovy lar- 

 vae, then larger larvae are likely to be much more 

 experienced with diverse attacks than are smaller 

 larvae. Response thresholds decreased linearly 

 with contacts up to 10 encounters in Dill's experi- 

 ments, but this effect would undoubtedly decline 

 with larger numbers of encounters, when a mini- 

 mum threshold would presumably be obtained. 

 Therefore, an age (experience) dependent decrease 

 in reaction distance may contribute to net avoid- 

 ance and the declining rate of catch probability at 

 larger sizes. 



148 



