samples collected different year classes, particularly in the "small" and "intermediate" size classes. However, stocking
rates were the same for all lakes (50/ha) and assuming relatively similar survival of stocked fish between years, I felt
that this would not compromise the study objectives.
Catch rates (CPUE) were expressed as number of fish per hour of electrofishing. Electrofishing catch rates
for each lake were compared by month and season, day and night, and habitat type for each size group. Catch data
from all lakes were not pooled due to differences in population abundance. CPUE data were log-transformed
[loge(CPUE-f 1)1 to normalize data and stabilize variances. Standard t-tests were used to determine differences in CPUE
by season and habitat type for each lake and size class. Paired t-tests were used to test for differences between day and
night samples collected from "saugeye" habitat. Ryan's multiple comparison test was used to determine differences in
CPUE by month. Statistical significance was assessed at P=0.05.
Sampling precision was measured by determining the coefficient of variation of the mean (CVx=S.E.x1). A
target level of precision was set at CV*=0.125. This value corresponds to the x+0.25x and coincides with standards
established for "management studies " by Robson and Regier (1964). Rearranging the above equation, inserting the
desired level of precision, and solving for N (number of samples) yields the equation:
N=0.125-2x-2s2. (equation 1)
Standard equations for estimating sampling size assume that the data are normally distributed and that the sample mean
and variance are uncorrelated. A mean-variance relationship for this study was calculated from all sampling strata, lakes,
and dates combined by linear regression of loges2 on loge*, yielding the equation:
loges2=1.61+1.221ogex. (equation 2)
The regression equation relating s2 and x was back-transformed to a linear scale and corrected for transformation bias
by adding the mean square error of the regression (MSE)/2. The mean-variance relationship for all samples collected
then becomes:
s2 =exp[(MSE/2) + 1.61 + 1.22x)] (equation 3)