Geoscience Reference
In-Depth Information
spatial pattern of the population of interest. Barabesi (2001) gave one of those
methods, applied to the estimation of plant density, based on (random) point-
to-plant distances. The procedure involves the estimation of the probability
density function of these distances through a kernel density estimator.
6.7 Computational Tools for Density
Estimation in Plotless Sampling
Equations entailing the estimation of densities using Byth's or any related esti-
mator as well as their corresponding standard errors in
T
T-square sampling are
simple as the
x
i
and
z
i
distances depicted in Figure 6.1 are the only informa-
tion needed. Any spreadsheet program or programming language like R can
be used for calculations. In fact, an R script was prepared for producing esti-
mates, standard errors, and plots (Figures 4.1 and 4.2) in the Lansing Woods
example (see the topic's companion site: https://sites.google.com/a/west-inc.
com/introduction-to-ecological-sampling-supplementary-materials/home/
chapter-6-r-code). As another option, the program Ecological Methodology
(Exeter Software, 2009), a companion software of Krebs's (1999) topic, also
offers Byth's and other estimators for
T
T-square sampling.
Regarding density estimation in wandering-quarter sampling, the basic pro-
cedure presented here can also be performed using any spreadsheet or pro-
gramming language. Generally, density estimates for any of the rapid plotless
sampling methods described in this chapter can be readily available to sur-
veyors in the field, given the versatile capacities of current portable computers.
However, it would be worth considering the development of special software
allowing the computation of estimators for a wider variety of assumptions in
plotless sampling methods (e.g., accounting for excessively clumped objects).
As an example, computational tools are needed for density estimation in wan-
dering-quarter sampling when objects are aggregated following the estimation
procedure originally proposed by Catana (1963). Finally, a suitable estimate of
the standard error of density in wandering-quarter sampling (e.g., by bootstrap-
ping) is waiting to be proposed, with an algorithm implemented in computers.
