the oribatid mite data, which will first be detrended (Sect. 7.3.2) to make them
second-order stationary (Sect. 7.2.6).
# The species data are first detrended; see Sect. 7.3
mite.h.det <- resid(lm(as.matrix(mite.h) ~ ., data = mite.xy))
mite.h.D1 <- dist(mite.h.det)
(mite.correlog
nperm = 999))
summary(mite.correlog)
# Plot the Mantel correlogram
plot(mite.correlog)
Hint In this run, the number of classes has been computed automatically using Sturge’s
rule. Use argument n.class to provide a user-determined number of classes.
In this simple run, most default settings have been applied, including Holm’s
correction for multiple testing (see Section 7.2.6). The number of classes has been
computed using Sturge’s rule: number of classes ¼ 1 + (3.3219 Â log 10 n), where n is
the number of elements, here the number of pairwise distances. The resulting
number of classes and the corresponding break points can be read in the result object:
# Number of classes
mite.correlog$n.class
# or: mite.correlog[2]
# Break points
mite.correlog$break.pts
# or: mite.correlog[3]
Hint The default option cutoff = TRUE limits the correlogram to the distance
classes including all points (the first 7 distance classes in this example); the
results for the last 5 distance classes (computed on fewer and fewer points) are
not shown.
The result shows significant positive spatial correlation in the first two distance
classes (i.e., between 0.15 m and 1.61 m; see the break points) and negative
significant correlation in the fourth to sixth classes (between 2.34 and 4.52 m).
Examining the environmental variables allows some speculation about the ecological reasons behind these structures. Close sites tend to show similar communities
because the soil conditions are rather similar. On the other hand, any pair of sites
whose members are about 2–4 m apart falls into contrasting soil conditions, which in
turn explains why their mite communities are different.
7.2 Spatial Structures and Spatial Analysis: A Short Overview
307
second-order stationary (Sect. 7.2.6).
# The species data are first detrended; see Sect. 7.3
mite.h.det <- resid(lm(as.matrix(mite.h) ~ ., data = mite.xy))
mite.h.D1 <- dist(mite.h.det)
(mite.correlog
nperm = 999))
summary(mite.correlog)
# Plot the Mantel correlogram
plot(mite.correlog)
Hint In this run, the number of classes has been computed automatically using Sturge’s
rule. Use argument n.class to provide a user-determined number of classes.
In this simple run, most default settings have been applied, including Holm’s
correction for multiple testing (see Section 7.2.6). The number of classes has been
computed using Sturge’s rule: number of classes ¼ 1 + (3.3219 Â log 10 n), where n is
the number of elements, here the number of pairwise distances. The resulting
number of classes and the corresponding break points can be read in the result object:
# Number of classes
mite.correlog$n.class
# or: mite.correlog[2]
# Break points
mite.correlog$break.pts
# or: mite.correlog[3]
Hint The default option cutoff = TRUE limits the correlogram to the distance
classes including all points (the first 7 distance classes in this example); the
results for the last 5 distance classes (computed on fewer and fewer points) are
not shown.
The result shows significant positive spatial correlation in the first two distance
classes (i.e., between 0.15 m and 1.61 m; see the break points) and negative
significant correlation in the fourth to sixth classes (between 2.34 and 4.52 m).
Examining the environmental variables allows some speculation about the ecological reasons behind these structures. Close sites tend to show similar communities
because the soil conditions are rather similar. On the other hand, any pair of sites
whose members are about 2–4 m apart falls into contrasting soil conditions, which in
turn explains why their mite communities are different.
7.2 Spatial Structures and Spatial Analysis: A Short Overview
307