Leis et al Behavioral ontogeny In larvae and early luvenile Coronx ignobilis 



405 



All bearings are given as degrees magnetic. 

 Cardinal directions are not abbreviated, but 

 other directions are abbreviated (e.g., SE for 

 southeast). From these bearings and the flow- 

 meter data, we estimated swimming direction 

 and speed in relation to the water (not the 

 bottom). 



Circular statistical procedures followed 

 Batschelet (1981) and Zar (1996). Mean vec- 

 tor length (r), is a measure of angular disper- 

 sion ranging from (maximum dispersion) to 

 1 (lack of dispersion). The Rayleigh test was 

 used for single-sample hypotheses about direc- 

 tional swimming. Watson-Williams test was 

 used for hypotheses about directional swim- 

 ming involving more than one sample. Most 

 of the circular statistical procedures were per- 

 formed with Oriana software (vers. 2, Kovach 

 Computing Services, Pentraeth, Wales, UK). 

 For all statistical tests, we report actual P 

 values whenever possible, but consider P<0.05 to con- 

 stitute a "significant" result. 



Results 



Larvae used in the critical speed trials were 21-30 days 

 old, and ranged from 8.0 to 16.5 mm SL. Larvae up to 

 18 mm were used in the in situ observations. All fins 

 were formed in the smallest individual and over the 

 size range studied, the larvae were very deep bodied, 

 primarily silver, and usually had six dark vertical bars 

 laterally. Although the specimens used on any day did 

 not constitute a random sample of the available sizes, 

 they could be used to obtain a rough estimate of the 

 growth rate over the experimental period. The range of 

 size-at-age on any day was as much as 6.5 mm. Average 

 growth over this period was 0.36 mm/d, and the slope of 

 the size versus age line was significantly different from 

 zero, but the relationship between size and age was 

 weak (size = 0.36 age + 2.87, P<0.001, r2 = o.28). Total 

 lateral area increased approximately as the square of 

 SL (TA = 0.51SL~°\ r2 = 0.93), whereas propulsive area 

 increased at a lower rate (PA = 0.34SL^^'^, r- = 0.94). 



Critical speed 



Critical speed was measured in 54 individuals in eight 

 batches of six over 10 days. Critical speed increased 

 with size from about 12 cm/s in the smallest individu- 

 als (ca. 8 mm) to about 40 cm/s in the largest (16.5 mm, 

 Fig. 2). For SL, a linear model provided the best fit for 

 this relationship (t/„,t= 2.72SL-9.55, P<0.001, r- = 0.51, 

 n = 54), but the difference between the linear model and 

 the others was small (r- for the other models ranged from 

 0.43 to 0.49). Both TA and PA gave somewhat better fits 

 than SL ([/„,t= 0.39TA + 5.25, /-2 = 0.58; U„^^= 0.22PA 

 -h5.73, r- = 0.55). Size was a much better predictor of 

 critical speed than was age {U^^^^=0.90age-0.22, P<0.01, 

 r2 = 0.13, K = 54). 



The fastest individual within each 1-mm increment of 

 size was considered the best performer, and these con- 

 stituted 16.6'7f of all larvae. These nine best perform- 

 ing larvae swam at 20 to 40 cm/s. The nine best per- 

 forming larvae had the same increase in performance 

 with size (2.7 cm/s per each mm of growth) as the "all- 

 individuals" line, but were 8 cm/s faster at all sizes 

 ([7^,„ = 2.69SL-1.46, P<0.001, r2 = 0.76). Therefore, both 

 average performers and "best performers" increased 

 speed at 2.7 cm/s per each mm of growh in SL. 



Scaled speed in terms of body lengths per second did 

 not increase significantly with size (P=0.07, r'~ = 0.06), 

 and the overall mean (±standard error) was 19 ±0.7 

 BL/s. The fastest 10% of individuals swam at a mean 

 t/,„, of 28.2 ±0.6 BL/s. 



In situ speed 



In situ speed was measured in 24 individuals ranging in 

 size from 8.5 to 18 mm SL. Some individuals stopped and 

 hovered at times during the observation period, but these 

 pauses in swimming were incorporated into the speeds 

 measured over the observation period. Observations were 

 made a week apart when the larvae were 24 and 31 days 

 old, respectively, and at a total of four locations that dif- 

 fered in depth, exposure, and bottom type. Observed in 

 situ speeds ranged from 3 to 19 cm/s. The in situ speeds 

 were less than the critical speed measurements (see 

 below). Mean scaled in situ speed was 9.5 ±0.9 BL/s. 



Overall, there was no significant increase in speed 

 with size (P=0.89, r-=0.001, n=24), but it was obvious 

 that the larvae behaved differently in the different loca- 

 tions; therefore we examined the data for each location 

 separately. Further, we eliminated observations on one 

 larva that simply drifted without swimming and on four 

 individuals that quickly swam to depths greater than 

 18 m and thus provided less than 5 minutes of observa- 

 tions, because short measurements of speed taken at 

 steep angles of descent can be inaccurate. The in situ 

 speed data on the 19 remaining individuals were used. 



