408 
Fishery Bulletin 95(3), 1997 
is not a reasonable sampling scheme for most fisher- 
ies but provides a standard against which the others 
may be judged. Many more abalone must be mea- 
sured under scenario 1 than scenario 2 to achieve 
comparable error indices (Figs. 4 and 5). For example, 
if under scenario 1 a total of 10 diver-days are 
sampled (approximately 3,400 abalone measured), 
then the error index is approximately 0.04. This level 
of error could be achieved by measuring only 400 
abalone randomly distributed across all diver-days. 
As an example of variation in sampling under sce- 
nario 2, consider the differences in error index and 
difference in mean size when sampling 100 abalone 
and 1,000 abalone in a fishery of 400 diver-days (Fig. 
5). Across the range considered ( 100-600 diver-days), 
the size of the fishery made little difference to esti- 
mates of error in the size composition or mean size 
of individuals (Fig. 5). 
Under scenario 3, the most realistic of the sam- 
pling schemes, there were large differences in the 
error index, depending both on the number of aba- 
lone sampled in total and the number of diver-days 
sampled (Fig. 6). The magnitude of error was not, 
however, greatly influenced by the total number of 
diver-days per year (Fig. 6). The results suggest that, 
although the error in estimating the size-frequency 
composition depended on the total number of aba- 
lone sampled, the rate of decline in the error index 
was similar among all sample sizes (Fig. 6). In all 
cases, the rate of decline in the error index reached 
an asymptote at approximately 20 diver-days. For 
zones or groups of zones with up to 600 diver-days 
per year, there was an approximate two-fold reduc- 
tion in the error index by increasing the total num- 
ber of abalone measured from 100 to 1,000 abalone 
(Fig. 6). There was little further reduction in the er- 
ror index by increasing replication from 
1,000 to 1,500 abalone (Fig. 6). 
The average error in estimated mean 
size of abalone declined rapidly with 
increasing number of diver-days be- 
tween 1 and 15 diver-days (Fig. 7). Fur- 
ther increases in simulated sampling 
effort produced relatively modest gains 
without large increases in the number 
of diver-days sampled. For example, 
doubling the sampling effort from 15 to 
30 diver-days caused only minor in- 
creases in precision (Fig. 7). In contrast 
to the patterns in errors in estimated 
size composition, increases in the total 
number of abalone measured from 100 
to 1,500 produced little reduction in the 
average difference in mean size (Fig. 7). 
The relative importance of diver-days 
as a source of variation is demonstrated 
by the difference between sampling a 
total of 100 abalone spread across five 
days and sampling 2 abalone on each of 
50 diver-days (Fig. 7). In the latter case, 
the average difference in mean size be- 
tween the true and estimated distribu- 
tions and the sample was approximately 
2 mm, in contrast to 6 mm when 100 
abalone were sampled in 5 diver-days. 
The difference between observed and ex- 
pected means was considerably smaller 
in comparing 200 abalone in each of 5 
diver-days with the same total number of 
abalone spread over 50 diver-days irre- 
spective of the size of the fishery (Fig. 7). 
In considering the sampling scheme 
required for the whole fishery, scenario 
Figure 4 
Errors in size composition and estimated mean size of abalone caught in 
different sizes of the “fishery” with sampling scenario 1. Calculation of the 
error index for the size composition and of the average difference in mean 
is described in the text. 
