Environmental Engineering Reference
In-Depth Information
probabilities as much as true abundance. An appropriate census method should
ideally be used in at least a sub-sample of sites in order to test whether there is
significant variation in detection probabilities between them (Section 2.3). For
example, in the case of presence-absence data,
occupancy models
can be used to
quantify variation in detectability with site characteristics (Royle, Nichols and
Kéry 2005, Section 2.3.6.2). Failure to account for detectability may seriously
undermine your study. For example, where hunting causes target species to become
more wary of people, uncritical use of an abundance index based on sightings will
overestimate the impact of hunting.
The choice of environmental variables to measure as possible
correlates of
abundance
is often huge, and some educated judgement is required in order to
identify a limited set of candidate variables that are both measurable and likely to
have an important influence. Ideally, a direct measure of
harvest effort
such as the
total number of harvesters active per unit area will be used, although proxies such
as distances to roads or settlements might be used if there is strong reason to believe
that they correlate well with harvest effort.
Habitat variables
might be broad
classifications of type, or focus more closely on aspects of the ecology of the focal
species such as the occurrence or availability of key resources and constraints. For
example, in the case of plants, soil type and moisture might be candidate variables,
while for an animal, one might look for factors related to the availability of food
and breeding sites.
Environmental variables can be measured
on the ground
, e.g. by visually classi-
fying habitats at survey sites, or by measuring quantifiable factors such as soil
fertility. In addition, many variables can be measured by
remote sensing
. For exam-
ple, altitude, distances to roads or settlements, and coverage by water might be
taken from topographic maps, while habitat type might be measured using aerial
photographs or remotely sensed satellite imagery. Spatial data of these kinds are
best manipulated in geographical information systems (GIS) in order to calculate
summary statistics such as the density of roads in each survey site. The benefit of
measuring potential correlates of abundance remotely is that they can be measured
across entire landscapes. Assuming that a model with good predictive power is
obtained from the sample of survey sites, this allows the focal species' abundance
across the landscape to be predicted without the need for further ground surveys.
Definition of sampling units
The results of spatial analysis are often sensitive to the
size of sampling unit
(Boyce
2006). For example, for a small plant whose abundance is largely determined by
local conditions on the scale of metres, variables measured at the scale of kilometres
will not pick up the relevant factors. Conversely, for a highly mobile animal, fine-
scale measures may fail to pick out any responses, or reflect small-scale movements
rather than large-scale habitat suitability. For the purposes of assessing harvest
impacts, the scale at which harvesters operate will also be relevant. Small-scale
harvest on foot will vary at the scale of one to a few kilometres, while large-scale
