Geoscience Reference
In-Depth Information
the changed region. Importantly, the conclusions are always specific to the
statistical population from which the sample was drawn; hence, changing
the population might change the results.
When defining the region of interest, some consideration should be given
to whether there are certain areas that might be regarded as nonhabitat
(i.e., no chance of the species being present there). If these areas are not of
interest, then they could be excluded and no sampling effort be allocated to
them. However, for longer-term studies or monitoring programs, if those
areas might potentially become habitat or become otherwise occupied in
the future, then some sampling effort should be devoted to those areas from
the outset such that future changes can be identified. Actual surveys for the
species might not have to be conducted from the outset, but those types of
places should be included as part of the overall sampling frame. Similarly,
if the intent is to identify changes in a species range over time, areas that
are outside the current range should be included in the sampling frame
to identify the species arriving at those locations. Absences should not be
regarded as no information as they define which areas are outside the spe-
cies current range.
9.6.2 Defining a Sampling Unit
As noted previously, the sampling unit is the item of interest for which
the occupancy category is to be determined. For example, at this scale the
intent might be simply to identify whether the species is present or absent.
Depending on the context, the units may be naturally (e.g., ponds or forest
patches) or arbitrarily (e.g., grid cells) defined, and it is the collection of all
such units that defines the population of interest.
What would be an appropriate definition will depend on the exact appli-
cation and the primary objective of the study. When interest is in a discrete
set of naturally occurring units, such as ponds, then the choice of sampling
unit will often be relatively straightforward. However, one additional con-
sideration is whether the units themselves will be dynamic through time.
For example, in a wetland complex or area of vernal pools, what may be con-
sidered a single pond in one year might be divided into multiple ponds the
next or vice versa. Thus, the very population of interest (in a statistical sense)
is changing each year, making it difficult to interpret any apparent changes
in the occupancy metric. In such situations, an arbitrarily defined sampling
unit might be more appropriate, with an initial consideration each year being
whether there is currently water, for example, in the unit.
Where there is no natural definition, careful consideration should be given
to the spatial extent to which an occupancy state is believed to apply. The
spatial extent of a sampling unit is often implicitly assumed, and by consid-
ering such issues prior to collecting the data, these assumptions are being
made explicit and can help formalize the sampling framework. For example,
when a species has been detected at a location, it is unlikely that a dot on a
