infoclust | R Documentation |
Function performs hierarchical clustering of binary ecological communities based on information analysis as defined by Williams et al. (1966) and Lance & Williams (1966).
infoclust(x, delta = TRUE)
x |
Community data |
delta |
Use increase in information ( |
Function performs information analysis of binary ecological communities (Williams et al. 1966, Lance & Williams 1966). The current implementation is based on Legendre & Legendre (2012).
The information I
of a collection of N
sampling units
with S
species is defined as
I = S N \log N - \sum_i^S a_i \log a_i + (N-a_i) \log (N-a_i)
where a
is the frequency count of each species in the
collection. The method works by merging either the units that give
the lowest increase (\Delta I
when delta = TRUE
), or
the units that are most homogeneous (lowest I
when
delta = FALSE
). After merging sampling units or clusters,
the community data matrix is updated by actually merging the data
units and re-evaluating their information distance to all other
units. The information content of all non-merged clusters is I
= 0
, and for clusters of several sampling units the constant
species (completely absent or always present) do not contribute to
the information. The largest increase in information is made by
species with 0.5 relative frequency, so that the analysis tries to
build clusters where species is either always present or always
absent. This often gives easily interpretable clusters.
Function returns an object of class "infoclust"
that
inherits from hclust
. It uses all "hclust"
methods, but some may fail or work in unexpected ways because the
analysis is not based on dissimilarities but on binary data matrix.
Williams, W.T., Lambert, J.M. & Lance, G.N. (1966). Multivariate methods in plant ecology. V. Similarity analyses and information-analysis. J. Ecol. 54, 427–445.
Lance, G.N. & Williams, W.T. (1966). Computer programs for hierarchical polythetic classification (“similarity analyses”). Comp. J. 9, 60–64.
Legendre, P. & Legendre, L. (2012). Numerical Ecology. 3rd English Ed., Elsevier.
## example used to demonstrate the calculation of
## information analysis by Legendre & Legendre (2012, p. 372).
data(pond)
cl <- infoclust(pond)
plot(cl, hang = -1)
## Lance & Williams suggest a limit below which clustering is
## insignificant and should not be interpreted
abline(h=qchisq(0.95, ncol(pond)), col=2)
## Spurn Point Scrub data
data(spurn)
cl <- infoclust(spurn)
plot(cl, hang = -1)
if (require(vegan)) {
tabasco(spurn, cl)
## apply information clustering on species
tabasco(spurn, cl, infoclust(t(spurn)))
}
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.