442 
Fishery Bulletin 96(3), 1 998 
Design of stock monitoring program 
Abundance estimates of abalone stocks in Victoria 
were made annually at 60 sites, using radial 
transects during 1992 to 1994. These estimates were 
made during routine stock surveys for the Victorian 
program for abalone stock assessment. At each site, 
nine replicate abalone collections were made. Sites 
were selected from abalone habitat with commercial 
quantities of abalone. During 1992, sites for radial 
transect surveys were initially chosen at random 
from between the 5-m and 18-m isobaths to avoid 
the wave-break zone and to minimize hyperbaric 
exposure. Accurate navigational fixes were made at 
each site with a global positioning system (GPS). 
During successive years each site was resurveyed 
after having been located with GPS. This mixed sam- 
pling design has been shown to have greater power 
than nested designs for benthic monitoring and 
avoids the potential bias of fixed designs (Van der 
Meer, 1997). 
Application of methods to stock monitoring 
An analysis of variance was applied to results from 
stock monitoring surveys to determine the inter- 
annual variation in abundance for each zone. This 
was in the form of a mixed linear model in the SAS® 
general linear model procedure. 
The relative treatment magnitude for each effect 
in the GLM was calculated with the same procedure 
as that used in the trial stock surveys. 
The effect of differences between research divers 
on abalone abundance estimates from radial 
transects was further investigated (with two- and 
three-way interactions omitted) across all zones from 
an analysis of 1992-94 stock surveys. During each 
field survey of a group of adjacent sampling sites, 
the dive team was composed of three or four out of a 
possible twelve divers. Differences in the composi- 
tion of dive teams between surveys occurred for lo- 
gistical reasons. At each site three divers performed 
three radial transects each where the allocation of 
directions for these transects among divers was ran- 
dom. Differences among divers were compared by 
using Ryan’s test (Day and Quinn, 1989), and the 
effect of diver experience on abundance estimates was 
determined by regressing divers’ mean collections of 
abalone per transect against the number of transects 
each diver performed. 
Power to detect changes in abundance 
Monte Carlo simulations were made to estimate the 
probability of detecting cumulative annual changes 
in abundance. The simulations were based on data 
from stock surveys conducted during 1992 and 1993 
that employed radial transects as well as data from 
1989 and 1991 surveys based on timed searches. Sce- 
narios involving different combinations of annual 
change increments (increases) and number of years 
sampled were each simulated 200 times for each 
management zone (i.e. central, eastern and western; 
Fig. 1). Divers were randomly selected from those 
who participated in the surveys, and the abundance 
estimates for each radial transect were adjusted for 
the diver factor and variation within each site, 
whereas the abundance estimates for timed searches 
were standardized by using a linear regression be- 
tween pairs of divers (McShane, 1994). The respec- 
tive analyses of variance used to analyze the stock 
survey data for each method were applied to each 
simulation, and the proportion of tests that showed 
a significant (a=0.05) annual change in abundance 
(year effect) was calculated. The proportion of sig- 
nificant results for each management zone was then 
plotted against the number of years simulated for each 
increment level. A smoothing function was applied to 
each curve to eliminate fluctuations due to changes 
in confidence levels among years in the series. 
Results 
Operator effects on methods 
Diver effects on radial transects During the first 
experiment at Gabo Island, the precision (SE/ic) of 
the divers’ abalone counts for each transect ranged 
from 0.04 to 0.27. Cochran’s test showed the divers’ 
variances of precision to be homoscedastic (a=0.01). 
Differences among divers’ mean precision values were 
not significantly different (F=1.43; df =3, 8; P>0.10). 
Collectively, the four groups of twelve estimates of 
abundance represent four identical surveys of the 
same site which yielded similar mean abundance 
estimates on each occasion (Fig. 2). Estimates of 
mean abalone abundance for the site during the sec- 
ond experiment were also similar (Fig. 3) and no sig- 
nificant differences were detected among the four 
divers (F=0.01; df=3, 33; P>0.10). 
Comparison of diver effects on radial transect and 
timed search survey methods There were no sig- 
nificant differences between the two divers when 
using radial transects (F=0.19; df=l, 2; P>0.10) nor 
were there significant differences when using timed 
searches (F=7.31; df=l, 2; P>0.10). Notably, the dif- 
ferences between the divers’ mean abundance esti- 
mates (Fig. 4) were much smaller for radial transects 
