Andrew and Chen: Estimating size structure and mean size of Haliotis rubra 
41 
provide reliable estimates of the size structure of 
abalone in the landed catch and would detect rela- 
tively small changes in the mean size of abalone at 
harvest for the whole fishery. How small those 
changes are will be determined by how likely man- 
agers are prepared to accept the possibility of being 
wrong. At a smaller scale, 1,000 abalone from 20 
diver-days would be needed to achieve a similar de- 
gree of discrimination. More intensive sampling did 
not greatly improve the reliability of these estimates 
over the range simulated. The simulations suggest that 
differences among diver-days were an important source 
of variation (see also Kitada et al., 1992; Crone, 1995). 
Interpretation of trends in mean size of individu- 
als among years requires some care. Divers may 
change their diving patterns among years, and fish 
populations may change with different demographic 
attributes. These patterns alone may produce 
changes in the composition of the landed catch inde- 
pendent of any underlying trend in the fishery. In 
essence, this problem is analogous to that in inter- 
preting catch-rate information from heterogenous 
fisheries, such as abalone. Several authors have 
claimed that apparent stability of catch rate is pos- 
sible despite declining abundance of abalone as a 
result of changing diver behavior (e.g. Hilborn and 
Walters, 1987). 
In these simulations we did not weight sampling 
effort in scenario 3 for the number of abalone caught 
in a diver-day. We, therefore, assume that there is 
no relation between the size and number of abalone 
caught. Using data from 1993-94, we found that 
there is no significant relation between the number 
of abalone caught in a diver-day and the size of those 
abalone. The impact of weighting is further dimin- 
ished by the fact that the fraction sampled per day is 
relatively high, irrespective of the number caught 
(usually >25%). We assume that this sampling frac- 
tion provides a reliable estimate of mean size within 
diver-days. 
The results of these simulations suggest that the 
large sample sizes possible in estimating mean size 
