Environmental Engineering Reference
In-Depth Information
UVC is a somewhat unusual application of plot sampling in that it is applied to
species that may be highly mobile. This can lead to heavily biased estimates, either
under-estimation caused by disturbance, or in some cases over-estimation caused
by attraction to the surveyor's activity (Edgar
et al
. 2004). In principle, more
sedentary fish species would be ideally suited to plot sampling, but these species are
often difficult to see, and it would probably be impossible to search the large plots
used in this example with sufficient confidence. Smaller plots searched more care-
fully would be appropriate in this case, but even then densities of species that hide
themselves in the substrate may be underestimated hugely (Willis 2001).
It also misses the non-territorial sector of the population, which can be significant,
both ecologically and in terms of harvest offtake. For example, Martin (1991)
found that when male willow ptarmigan
Lagopus lagopus
were removed from their
territories, 70% of them were quickly replaced by unpaired males.
This method is aimed at
static species
, particularly plants. It works by randomly
selecting a number of individuals or points within the study area, and measuring
the distance to the nearest neighbouring individual. This is actually conceptually
similar to plot sampling, but with a fixed number of objects per plot and variable
plot size rather than
vice versa
. Given a sample of
n
distances,
d
i
, in a study site of
area
A
, the population estimate is:
nA
N
ˆ
n
d
i
i
1
Confidence intervals for the estimate can be calculated using the non-paramet-
ric bootstrapping procedure described in the previous section. For further details
see Borchers
et al
. (2002).
While this technique is appealingly simple, it has a major flaw, in that it gives
seriously biased estimates if the study subjects are not randomly and independently
distributed. Because this is rarely true in practice, plotless sampling is very rarely a
reliable method. Trying to locate nearest neighbours can also be surprisingly time
consuming.
When searching for any object, the further it is from you, the less likely you are to
see it. Distance sampling works by using survey data to quantify this decline in the
probability of detection
, which then allows you to infer how many objects were
there but not seen. It works like this. While searching for your target species,
instead of pre-defining a fixed area within which you assume everything will be
