ordtest: Ordination Distribution Test

Description

Testing the distribution of points in an ordination

Usage

ordtest(ord, var, dim=1:ncol(ord$points), index = 'euclidean',
   nitr = 1000)

Arguments

ord

an object of class ‘pca’, ‘pco’, or ‘nmds’

var

a logical or factor vector used to organize the calculation of within-set distances

dim

the number of dimensions to use in the calculation

index

the distance metric for the calculation of within-set distances. Currently only euclidean is accepted

nitr

the number of iterations to perform to establish p-values

Value

Produces a list with components:

obs

the observed sum of within-set distances

the probability of obtaining a value that small

reps

the sum of within-set pairwise distances for all permutations

Details

Calculates the sum of within-set pair-wise distances and compares to ‘nitr’ permutations of the same distribution to calculate the probability of observing clusters as tight as observed or tighter. The p-value is calculated by running nitr-1 permutations and counting the number of cases where the sum of pair-wise distances is as small as smaller than observed. That count is increased by one and divided by nitr to estimate p.

Examples

Run this code

# NOT RUN {
data(bryceveg)
data(brycesite)
dis.bc <- dsvdis(bryceveg,'bray/curtis')
pco.bc <- pco(dis.bc)
plot(pco.bc)
demo <- ordtest(pco.bc,brycesite$quad)
demo$p
# }