The 2007 survey was stratified on the basis of four toheroa bed strata (strata 1 to 4), and one non-bed
stratum (stratum 5) (Akroyd et al. 2008). In our reanalysis of those survey data, for each stratum, the
original transect data were resampled with replacement 1000 times (generating a new set of data with
the same number of transects as the original survey), and the mean number of adult toheroa
calculated. In addition, similar resampling with replacement was conducted with the same sample
size, but randomly selecting quadrats at 10 m intervals (starting at the first or second quadrat from the
high water mark), and repeating this approach with 1.75 times the original 5 m transects per stratum
(roughly the equivalent sampling effort, allowing for travel time). The distribution of estimates of
mean toheroa per transect, median, and 5th to 95th percentile range for each of our three resampling
approaches are shown in Figure 5. For stratum 3 in particular (which had the highest density in 2007
at approximately 1100 toheroa per transect) there was a marked decrease in precision (increase in the
5th to 95th percentile range) with 10 m quadrat spacing, even with increased numbers of transects
sampled. Strata 1, 2, and 4 all had similar mean numbers of toheroa per transect (approximately 500)
but gave contrasting results, with a drop in precision with 10 m spacing in stratum 2 versus no real
difference in precision with 5 m or 10 m spacing in strata 1 and 4. Our simulation suggested that
sampling at 5 m spacing of quadrats may be of real benefit (in terms of increased precision) for
transects that traverse high density toheroa beds (more than 500 toheroa per transect).