The two data sets used in most of this book do not lend themselves to a space-time
interaction analysis. Therefore, we will use a dataset that is provided in the
adespatial package, and is also used as an example in the Legendre et al.
(2010) paper. It consists in the counts over 10 periods of 10 days of the abundances
of 56 adult Trichoptera (Insecta) species emerging in 22 emergence traps laid along
the outflow stream of Lac Cromwell on the territory of the Station de Biologie des
Laurentides (Université de Montréal). We have thus a linear spatial layout of
22 points and 10 time points, yielding 220 observations.
Among the two functions to perform space-time interaction analysis,
stimodels()offers more options, and quicksti() provides a quick way to
run an sti analysis under Model 5 followed by tests of the main factors. The latter are
run under Models 6a and 6b if the interaction is significant, and under Model 5 (or 2:
standard test of the main factors using their Helmert contrasts) otherwise. Let us
compute an sti analysis with the function stimodels().
Beware: the spatial and temporal eigenfunctions constructed by the original
functions of the STI package (distributed as a Supplement to the paper) are actually
first-generation PCNMs. By default the functions retained half of the PCNM variables, which, at least in case of regular sampling designs, correspond to positive
autocorrelation. The stimodels() and quicksti() functions incorporated in
package adespatial compute dbMEM functions, retaining those with positive
spatial autocorrelation.
# Load the Trichoptera data
data(trichoptera)
names(trichoptera) # Species codes
The first two columns contain the site and date numbers, respectively.
# Hellinger transformation
tricho.hel <- decostand(trichoptera[ ,-c(1,2)], "hel")
# sti analysis
stimodels(tricho.hel, S = 22, Ti = 10, model = "5")
It is useless to store the results into an object unless one wants to use the function
in a workflow involving the retrieval of specific sti results; the output object only
contains the numerical results of the tests in a raw form. The analysis, run as
described above, produces the following on-screen display:
7.6 Space-Time Interaction Test in Multivariate ANOVA, Without Replicates
365
interaction analysis. Therefore, we will use a dataset that is provided in the
adespatial package, and is also used as an example in the Legendre et al.
(2010) paper. It consists in the counts over 10 periods of 10 days of the abundances
of 56 adult Trichoptera (Insecta) species emerging in 22 emergence traps laid along
the outflow stream of Lac Cromwell on the territory of the Station de Biologie des
Laurentides (Université de Montréal). We have thus a linear spatial layout of
22 points and 10 time points, yielding 220 observations.
Among the two functions to perform space-time interaction analysis,
stimodels()offers more options, and quicksti() provides a quick way to
run an sti analysis under Model 5 followed by tests of the main factors. The latter are
run under Models 6a and 6b if the interaction is significant, and under Model 5 (or 2:
standard test of the main factors using their Helmert contrasts) otherwise. Let us
compute an sti analysis with the function stimodels().
Beware: the spatial and temporal eigenfunctions constructed by the original
functions of the STI package (distributed as a Supplement to the paper) are actually
first-generation PCNMs. By default the functions retained half of the PCNM variables, which, at least in case of regular sampling designs, correspond to positive
autocorrelation. The stimodels() and quicksti() functions incorporated in
package adespatial compute dbMEM functions, retaining those with positive
spatial autocorrelation.
# Load the Trichoptera data
data(trichoptera)
names(trichoptera) # Species codes
The first two columns contain the site and date numbers, respectively.
# Hellinger transformation
tricho.hel <- decostand(trichoptera[ ,-c(1,2)], "hel")
# sti analysis
stimodels(tricho.hel, S = 22, Ti = 10, model = "5")
It is useless to store the results into an object unless one wants to use the function
in a workflow involving the retrieval of specific sti results; the output object only
contains the numerical results of the tests in a raw form. The analysis, run as
described above, produces the following on-screen display:
7.6 Space-Time Interaction Test in Multivariate ANOVA, Without Replicates
365