Environmental Engineering Reference
In-Depth Information
methods, followed by some guidance on how to select the most appropriate one in
your case.
Plot sampling refers to any form of fixed area that is searched for the target species.
Searched areas may be square quadrants, rectangular strips or belts, round plots
or even irregularly shaped regions defined by landscape features if this is most
convenient, although extra care needs to be taken to avoid bias in this case. Plot
sampling is the most basic analytical method, requiring no estimation of
detectability because we assume that all individuals within the plots are found.
Given a total of
n
individuals seen in
k
plots, each of area
a
, the estimated density
is therefore simply:
n
ka
ˆ
Assuming that individuals are randomly and independently distributed, and
that sample plots do not overlap, the population variance,
s
2
, can be used to mea-
sure the precision of the density estimate (see Box 2.1). If the population is in a
well-defined area of size
A
, the population size estimate is simply:
N
ˆ
ˆ
A
,
SE(
N
ˆ
)
SE(
ˆ
)
A.
However, if there is no overlap between plots (sampling without replacement), this
approach increasingly overestimates variance as the proportion of the whole area
covered,
p
, increases. In this case:
SE(
N
ˆ
)
SE(
ˆ
)
A
1
p
so that as proportional coverage approaches 1, the variance approaches 0. This
makes sense, because if we observe the entire population, we are no longer sam-
pling, and have complete certainty in the population count. The assumption that
individuals are independently and randomly distributed, required by the above
variance estimators, will frequently be violated in practice. More often, organisms
will be more clumped than expected, leading to under-estimation of variance
under the above approximations. In this case
non-parametric bootstrapping
may
be more appropriate (see Box 2.1).
The requirement that all individuals in a plot should be found generally restricts
plot sampling to
conspicuous species
that don't hide or flee from people. This
typically means sessile species, particularly plants; however, the method can occa-
sionally be applied to more mobile species (e.g. primates in small forest patches;
Mbora and Meikle 2004; Anderson
et al
. 2007; Box 2.2). For territorial animals,
repeat visits over time can be used to
map territories
, and so arrive at a total count
of territorial individuals (Bibby
et al
. 1992; Fuller
et al
. 2001). However, territory
mapping is very time consuming, and relies on territorial signals such as calls.
