Welsford and Lyle. Estimates of growth of Notolabrus fuacola from length- and age-based models 



703 



LB 



PB 



to (cm) 



L (cm) 



k (/year) 



Figure 2 



Bootstrap parameter estimates for Notolabrus fucicola, by site, for the standard von Bertalanffy growth function. Note 

 /, and k are plotted on logarithmic axes for clarity: (A) ! v vs. k (B) /, vs. t Q (C) k vs. t a . Contrasting crosses show the 

 location of parameter estimates based on the original data set (+, PB = Point Bailey, x, LB= Lord's Bluff). 



were more or less evenly distributed around the origi- 

 nal parameter estimates, resulting in approximately 

 symmetrical first-order corrected 95% CIs (Table 7). 

 Based on the lack of overlap of CIs, only g., differed 

 significantly between sites. A randomization test of the 

 difference in g 20 produced CIs of 0.75-2.85 cm/yr faster 

 growth at Lord's Bluff. 



Plots of bootstrap parameter estimates clearly indi- 

 cate differences in growth rates between sites, and little 

 overlap in the parameter clouds along the g 20 axis when 

 g 20 is plotted against g 30 (Fig. 4A). Plots of bootstrapped 

 estimates of the seasonal growth parameters u and w 

 showed a high level of nonlinear correlation. A region 

 of overlap between site estimates along the w axis is 

 evident in Fig. 4B. However, the randomization test for 



this parameter produced a CI of the difference between 

 the two sites of 0.02-0.33 yr, corresponding to signifi- 

 cantly different maximum in seasonal growth occurring 

 at Lord's Bluff 8-120 days after Point Bailey. Estimates 

 of w at Point Bailey ranged from -0.14 to 0.05 years in 

 relation to 1 January (Table 7), corresponding to peak 

 growth between austral mid-spring and mid-summer 

 (early November through mid-January), contrasting 

 with the Lord's Bluff estimate of -0.08 to 0.20 years 

 and indicating peak growth from austral late spring to 

 early autumn (mid-December through mid-Marchl. 



Site differences in growth were also indicated in the 

 results of LRTs. The overall models were significantly 

 different; the growth parameter g 20 and the timing of 

 maximum seasonal growth were significantly different 



