Haddon et al.: Using an inverse-logistic model to describe growth increments of Hahotis rubra 
65 
for the smallest abalone and reaching a maximum and 
tailing off as initial length increases. Finally, the inverse- 
logistic curve predicted constant increments for smaller 
abalone (initial linear growth) until the growth incre- 
ments began to decrease with increasing initial length. 
The minimum AIC and BIC were produced by the 
4-parameter inverse-logistic curve and not by the 3- 
parameter von Bertalanffy and Gompertz curves. The 
log-likelihoods were -726.1 for the von Bertalanffy, 
-712.2 for the Gompertz, and -681.1 for the inverse- 
logistic model. With only one more parameter than the 
other two models, a likelihood ratio test implies that 
the inverse-logistic curve was a significant improvement 
over the other two curves. 
Annual growth descriptions 
Within each of the three regional groups of sites, the 
predicted mean growth increment for given initial shell 
lengths for the 4-, 5-, and 6-parameter models was very 
similar (Fig. 4). In the case of the southwest and Actaeon 
regions, the predicted lines were visually coincident, 
whereas for Bruny Island there were only very slight 
differences in the three curves (Table 2). 
For both the southwest and the Actaeon regions, the 
4-parameter model was deemed the optimum model 
configuration by both the AIC and BIC. For the Bruny 
Island region, the BIC indicated that the 4-parameter 
model was optimal and the AIC indicated the 6-param- 
