Thanks to the introduction of the computation of R
2
adj in CCA, we could also
compute a variation partitioning based on CCA. However, we would have to do it
step by step or write a function, since no ready-made function is provided in vegan.
Interested readers are encouraged to write this function as an exercise.
6.4.2.4 Three-Dimensional Interactive Plots
Instead of plotting these parsimonious results as we did before, let us explore a
vegan function that can be very useful either for a researcher looking for a new
perspective on his or her results, or for a teacher: a 3D interactive plot. We will see
several options to reveal or combine results in different ways. These 3D plots are run
under the vegan3d package.
# Plot of the sites only (wa scores)
ordirgl(spe.cca.pars, type = "t", scaling = 1)
Using the mouse, enlarge the plot by dragging its lower right-hand corner. Then
move the plot around by left-clicking on any point in the plot with the left button.
Use the scroll wheel to zoom in and out.
# Connect weighted average scores to linear combination scores
orglspider(spe.cca.pars, scaling = 1, col = "purple")
The purple connections show how well the CCA model fits the data. The shorter
the connections, the better the fit.
# Plot the sites (wa scores) with a clustering result
# Colour sites according to cluster membership
gr <- cutree(hclust(vegdist(spe.hel, "euc"), "ward.D2"), 4)
ordirgl(spe.cca.pars,
type = "t",
scaling = 1,
ax.col = "black",
col = gr + 1
)
# Connect sites to cluster centroids
orglspider(spe.cca.pars, gr, scaling = 1)
262
6 Canonical Ordination
2
adj in CCA, we could also
compute a variation partitioning based on CCA. However, we would have to do it
step by step or write a function, since no ready-made function is provided in vegan.
Interested readers are encouraged to write this function as an exercise.
6.4.2.4 Three-Dimensional Interactive Plots
Instead of plotting these parsimonious results as we did before, let us explore a
vegan function that can be very useful either for a researcher looking for a new
perspective on his or her results, or for a teacher: a 3D interactive plot. We will see
several options to reveal or combine results in different ways. These 3D plots are run
under the vegan3d package.
# Plot of the sites only (wa scores)
ordirgl(spe.cca.pars, type = "t", scaling = 1)
Using the mouse, enlarge the plot by dragging its lower right-hand corner. Then
move the plot around by left-clicking on any point in the plot with the left button.
Use the scroll wheel to zoom in and out.
# Connect weighted average scores to linear combination scores
orglspider(spe.cca.pars, scaling = 1, col = "purple")
The purple connections show how well the CCA model fits the data. The shorter
the connections, the better the fit.
# Plot the sites (wa scores) with a clustering result
# Colour sites according to cluster membership
gr <- cutree(hclust(vegdist(spe.hel, "euc"), "ward.D2"), 4)
ordirgl(spe.cca.pars,
type = "t",
scaling = 1,
ax.col = "black",
col = gr + 1
)
# Connect sites to cluster centroids
orglspider(spe.cca.pars, gr, scaling = 1)
262
6 Canonical Ordination