Leis et al.: Behavioral ontogeny in larvae and early luvenile Caronx ignobilis 



409 



West Coast 



270 :? — IV 



B 



Nan Wan Bay 







Figure 7 



Frequency distributions of mean in situ swimming directions for giant trevally iCaranx 

 ignobilis) larvae at two locations. Thick lines represent mean values for individuals, and 

 the radius (thin line) that penetrates the outer circle represents the overall mean. The 

 numbers on dashed lines represent the number of times that bearing was recorded for a 

 larva. (A) West coast off Wan Li Tong and Her Chen, where larvae swam offshore: overall 

 mean=268°, r=0.59, P=0.04. (B) Nan Wan Bay, where larvae had no overall swimming 

 direction: overall mean = 146°, r=0.'29, P=0.58. "r" is the length of the mean vector, P is for 

 the Rayleigh test. 



Discussion 



Previous studies of the behavior of larval carangids 

 have focused primarily on their vertical distribution in 

 the ocean and have been based on plankton net samples 

 and on observations of the development of schooling- 

 related behavior in the laboratory (see references at the 

 beginning of this article). There have been no previous 

 studies on the ontogeny of orientation or of vertical dis- 

 tribution determined from in situ observations, and only 

 one article has considered the development of swimming 

 abilities of carangid larvae. 



Growth rates in C. ignobilis larvae of 0.36 mm/d in 

 the laboratory and ca. 0.3-0.8 mm/d at the aquaculture 

 farm at Kaohsiung, where we obtained larvae for our 

 experiment, were comparable to the few reported field 

 growth rates for larvae of other carangids. Small (<5 

 mm SL) larvae of Atlantic bumper (Chloroscombrus 

 chrysurus) grew at 0.3-0.4 mm/d in the Gulf of Mexico 

 (Leffler and Shaw, 1992), and larvae of two species of 

 Trachurus grew at 0.2-0.7 mm/d (Hewett et al., 1985; 

 Jordan, 1994). The C chrysurus larvae were found at 

 26-30°C, whereas the two Trachurus spp. were living 

 at about 15°C. In T. declivis (greenback horse mackerel), 

 8-mm larvae were about 17 days old and 12-mm larvae 

 were about 22 days old (Jordan, 1994). In T. syminetri- 

 cus (Pacific jack mackerel), a 10-mm larva was 40 days 

 old and a 20-mm larva was about 57 days old (Hewitt, 

 1981). Growth rates for C. ignobilis in culture or in the 



laboratory are unlikely to be relevant to field situations, 

 and the available field measurements of other species 

 are either for much smaller larvae or for individuals in 

 much cooler water and are unlikely to be applicable to 

 C ignobilis. Therefore, it is not possible to relate the 

 size-based performance measures reported in our study 

 to ages of larvae in the field. Further, for the labora- 

 tory-reared larvae that we used, size was a much better 

 predictor of performance than was age, as is generally 

 found to be the case (e.g., Fuiman and Higgs, 1997; 

 Clark et al., 2005). We found that using PA instead of 

 SL as a measure of size resulted in little or no improve- 

 ment in the proportion of the variation in swimming 

 performance that was explained by the linear relation- 

 ships developed in our study. Other studies have had 

 mixed results in this regard (Fisher et al., 2000; Clark 

 et al., 2005). 



Swimming speed of C. ignobilis increased rapidly 

 with growth for both critical speed and in situ speed, 

 although the rate of increase was greater for critical 

 speed. Both in terms of actual speeds and rate of in- 

 crease with growth, critical speeds of C. ignobilis are 

 within the range of reported values for larvae of other 

 warm water marine perciform fishes (Fisher et al., 

 2000; Clark et al., 2005). For larvae of benthic fishes 

 about to settle (ca. 8-12 mm SL), pomacentrids have 

 higher f/^^j, values than C. ignobilis, whereas apogonids, 

 percichthyids, sciaenids, and sparids have similar {/^.^jj 

 values. Seriola quinqueradiata is the only other caran- 



