Environmental Engineering Reference
In-Depth Information
●
Increase precision by increasing the
sample size
. Bear in mind, though, that
sampling error decreases with the square root of sample size (Box 2.1), giving
diminishing returns on effort.
●
Don't artificially improve precision through
pseudoreplication
(treating non-
independent sampling units as if they were independent). For example, this
might arise if large sampling units are arbitrarily subdivided after the event in
order to boost sample size. Occasionally, when surveying sparsely distributed
but locally common species, it may be cost effective to increase sampling
effort when aggregations are encountered, in which case
adaptive cluster
sampling
techniques can be used (Thompson 2002, 2004). Usually, though,
it is preferable to design the field study to deliver fully independent replicates.
●
Consider using a
stratified sample
. This is the purposeful partitioning of
sampling units between areas or groups that differ in ways thought likely to
influence the parameter of interest. For example, samples may be stratified by
habitat. This has the dual benefit of
improving precision
by accounting for
some of the variability between samples (although this only works well when
the number of strata is small relative to the total number of samples) and alle-
viating potential
problems with bias
due to under-sampling of less accessible
areas (so long as at least some sampling can take place in these areas).
●
Specific
environmental
variables can be measured at each sample site, and a
statistical model used to relate spatial variation in the parameter of interest to
these variables. As well as helping to increase precision, this approach can
provide information about the correlates and potential causes of variation,
helping to tease out harvest effects from other potential influences (see
Section 2.5.2 and Box 4.4).
●
When surveys are primarily aimed at measuring changes over time, precision
can be improved by
resampling
the same units on each occasion (Plumptre
2000). However, care must be taken with this approach to ensure that the
survey activity itself does not affect the parameter being measured. For
example, multiple visits to a plot may alter population size through habitat
disturbance.
Comprehensive coverage of sampling theory can be found in Thompson (2002).
Online resources providing more details on the subject are suggested in Section 2.7.1.
Abundance is a general term for the number of individuals in a population, which
can be measured either as population size or, more usefully, as density of individuals
within a given area. Knowing how to proceed in the field depends on knowing how
you are going to analyse the data you collect. It is therefore essential to know what
analysis methods are available to you, and how they work, so that the data can be
gathered in an appropriate fashion, minimising the chances of nasty surprises at
the analysis stage. This section provides an introduction to the available statistical
