Leis et al Behavioral ontogeny In larvae and early juvenile Caranx ignobilis 



407 



at about 5 minutes. The smaller (8-10 mm) 

 larvae then stayed at about 9 m, whereas the 

 larger (10-12 mm) larvae gradually ascended, 

 reaching a mean depth of about 6 m by the 

 end of the 10-min observation period. In con- 

 trast, larvae of 12-14 mm were more varied in 

 their depth ranges, some reaching the surface, 

 whereas others remained within 2-3 m of 

 the release depth (Fig. 5C). Larvae of 12-14 

 mm had no overall temporal trend in depth 

 and maintained a mean depth of about 5-6 

 m throughout the study period. Several of the 

 larvae of this group did oscillate over a depth 

 range of several m during the 10-min observa- 

 tion period. The largest larvae (14-18 mm) had 

 more varied vertical-distribution ranges, lead- 

 ing to high amplitudes (Fig. 5D). Three of the 

 five largest larvae swam deep monotonically 

 and the other two oscillated — one reaching 

 the surface after an initial descent, whereas 

 the other reached our lower dive limit ( 16 m), 

 but only after first ascending above the release 

 depth for the first 5 minutes. 



We found a weak relationship between 

 depth amplitude (difference between deep- 

 est and shallowest observations) and size 

 (a?np.=0A4SL + 1.7, P=0.03, r- =0.21, n=24); 

 larger individuals swam over a greater range 

 of depths owing to the tendency of larger lar- 

 vae to either oscillate or to swim deep mono- 

 tonically. Water column depth did not influ- 

 ence either mean swimming depth (P=0.11, 

 r2=0.11) or amplitude (P=0.06, r- = 0.15). 



Orientation 



Sixteen of the 24 larvae observed /?; situ had directional 

 trajectories (P<0.05, Rayleigh test, Table 3). Figure 

 6 represents the direction frequency distribution of 

 a C. ignobilis larva with average directionality. The 

 remaining analyses included only the 16 individuals 

 for which directional trajectories were found (termed 

 "directional larvae" or "directional individuals'" for the 

 sake of brevity), but the same result was obtained when 

 all 24 individuals were included. If all locations were 

 considered together, the mean swimming directions 

 of these 16 individuals had no overall directionality 

 (P>0.20, Rayleigh test). In contrast, there was signifi- 

 cant overall directionality off the west coast (Fig. 7A) 

 where the larvae swam, on average, offshore or toward 

 the west (269°, Rayleigh test, P=0.024). There was no 

 indication of any overall directionality in mean swim- 

 ming direction in Nan Wan Bay (Fig. 7B) (mean direc- 

 tion, 145°, P=0.58, Rayleigh test). Overall swimming 

 direction, however, differed significantly between the 

 west coast and Nan Wan Bay (Watson-Williams test, 

 P=0.028). 



We found only a limited indication that the direction- 

 ality of the larvae changed with size (Table 3). There 

 was no ontogenetic increase in precision of directionality 



(measured as r, the length of the mean vector) (r-=0.05, 

 P>0.10). Similarly, although the mean size of the eight 

 nondirectional larvae was 11.6 mm, and that of the 16 

 directional larvae was 12.2 mm, this difference was not 

 significant (^test, P=0.22). In terms of overall orienta- 

 tion, larvae observed off the west coast were smaller by 

 about 4 mm (^test, P=0.002) and were studied a week 

 earlier than those studied in Nan Wan Bay; therefore 

 it is possible that the difference in overall swimming 

 direction between locations was due to temporal or 

 ontogenetic factors rather than to spatial factors. At 

 the west coast location, the overall mean direction of 

 the five small (8-9.5 mm) directional larvae was the 

 same (268°) as the overall mean direction of the 4 large 

 (10-14.5 mm) directional larvae, indicating there was 

 no ontogenetic change in directionality. 



Finally, within trajectories, there was no increase in 

 directionality with time. The r value for the first half of 

 the trajectories was not significantly different from the 

 r value for the second half of the trajectory (Wilcoxon 

 signed-rank test, P>0.2). Nor was there any indication 

 that larger larvae had a greater difference in r between 

 the first and second halves of their trajectories than did 

 smaller larvae (r2=0.02, P»0.20). 



