FISHERY BULLETIN: VOL, 87, NO. 1 



Individual variability in rates of larval growth was 

 assessed by examining the distribution of residuals 

 around the mean exponential growth trajectory for 

 each population each year; a positive residual in- 

 dicates growth faster than average for the popula- 

 tion and a negative one growth slower than average. 

 Analysis of these residuals indicated that rates of 

 larval growth varied seasonally in both years (Fig. 

 10). Although the variability of rates of larval 

 growth was high within any given period, in both 

 years growth residuals differed significantly for lar- 

 vae hatched in different months (ANOVA F5 no = 

 6.72, P < 0.001, for 1984, and i^ggei = 50.86, P < 

 0.001, for 1985). In 1984, there was a weak, but con- 

 sistent tendency for residuals to increase through- 

 out the spawning season (correlation between resid- 

 ual and hatching date, r = 0.36, P < 0.01). In 1985, 

 deviations from population mean growth rates were 



generally negative early in the spawning season, 

 reached a positive maximum during August, and 

 then decHned in September. 



There was also evidence of a complex relationship 

 between rates of larval growth and location. Over- 

 all, the distributions of growth residuals differed 

 significantly across transects (i^gsjs = 8.71, P < 

 0.01), with relative growth rates tending to be 

 highest farthest from the west coast spawning area. 

 The weakness of the correlation between growth 

 rate and distance is due, in part, to two factors. 

 First, there was a marked change in the relation- 

 ship between location and growth rate with increas- 

 ing age of the larvae examined. The older the lar- 

 vae, the more positive the slope between distance 

 from the spawning area and relative growth rate 

 (Fig. 11). For larvae less than approximately 10 d 

 postfirst- feeding, the slope was significantly 



< 



C 



s 



15 

 3 



en 



o 



1 June 



1 July 



1 Aug 



1 Sept 



1 Oct 



1 July 



1 Aug 

 Spawning Date 



1 Sept 



Figure 10,— Temporal variation of residuals from the semilog regression of In total length 

 against age for 1984 and 1985, Macruronus novaezelandiae spawning started approx- 

 imately a month later in 1985 than 1984 (see Gunn et al. in press tor details). Differences 

 in residuals for larvae pooled by month of spawning are significant at P < 0.01 for both 

 years. 



40 



