vegan3d (version 0.65-1)

ordiplot3d: Three-Dimensional Ordination Graphics

Description

Function ordiplot3d displays three-dimensional ordination graphics using scatterplot3d. Function works with all ordination results form vegan and all ordination results known by scores function.

Usage

ordiplot3d(object, display = "sites", choices = 1:3, ax.col = 2,
        arr.len = 0.1, arr.col = 4, envfit, xlab, ylab, zlab, ...)

Arguments

object

An ordination result or any object known by scores.

display

Display "sites" or "species" or other ordination object recognized by scores.

choices

Selected three axes.

ax.col

Axis colour (concerns only the crossed axes through the origin).

arr.len

'Length' (width) of arrow head passed to arrows function.

arr.col

Colour of biplot arrows and centroids of environmental variables.

envfit

Fitted environmental variables from envfit displayed in the graph.

xlab, ylab, zlab

Axis labels passed to scatterplot3d. If missing, labels are taken from the ordination result. Set to NA to suppress labels.

...

Other parameters passed to graphical functions.

Value

• Function ordiplot3d returns invisibly an object of class "ordiplot3d" inheriting from ordiplot. The return object will contain the coordinates projected onto two dimensions for points, and the projected coordinates of origin, and possibly the projected coordinates of the heads of arrows and centroids of environmental variables. Functions like identify.ordiplot, points.ordiplot, text.ordiplot can use this result, as well as ordihull and other functions documented with the latter. The result will also contain the object returned by scatterplot3d, including function xyz.convert which projects three-dimensional coordinates onto the plane used in the current plot (see Examples). In addition, there is a function envfit.convert that projects a three-dimensional envfit object to the current plot.

Warning

Function ordiplot3d is based on scatterplot3d which does not allow exactly setting equal aspect ratio for axes. The function tries to circumvent this by setting equal plotting ranges for all axes so that the plot should be a cube. Depending on the dimensions of plotting device, this may fail, and the user should verify that the axes are approximately equal.

Please note that scatterplot3d sets internally some graphical parameters (such as mar for margins) and does not honour default settings. It is advisable to study carefully the documentation and examples of scatterplot3d.

Details

Function ordiplot3d plots static three-dimensional scatter diagrams using scatterplot3d. Function uses most default settings of underlying graphical functions, and you must consult their help pages to change graphics to suit your taste (see scatterplot3d). Function will display only one selected set of scores, typically either "sites" or "species", but for instance cca also has "lc" scores. In constrained ordination (cca, rda, capscale), biplot arrows and centroids are always displayed similarly as in two-dimensional plotting function plot.cca. Alternatively, it is possible to display fitted environmental vectors or class centroids from envfit. These are displayed similarly as the results of constrained ordination, and they can be shown only for non-constrained ordination. The user must remember to specify at least three axes in envfit if the results are used with these functions.

Function ordiplot3d plots only points. However, it returns invisibly an object inheriting from ordiplot so that you can use identify.ordiplot to identify "points" or "arrows". The underlying scatterplot3d function accepts type = "n" so that only the axes, biplot arrows and centroids of environmental variables will be plotted, and the ordination scores can be added with text.ordiplot or points.ordiplot. Further, you can use any functions from the ordihull family with the invisible result of ordiplot3d, but you must remember to specify the display as "points" or "arrows". To change the viewing angle, orientation etc.{} you must see scatterplot3d. Only one kind of scores will be plotted. See Examples for plotting both species and site scores.

See Also

scatterplot3d, ordiplot, identify.ordiplot, text.ordiplot, points.ordiplot, ordihull, plot.cca, envfit.

Examples

### Default 'ordiplot3d'
data(dune, dune.env)
ord <- cca(dune ~ A1 + Moisture, dune.env)
ordiplot3d(ord)
### A boxed 'pin' version
ordiplot3d(ord, type = "h")
### More user control
pl <- ordiplot3d(ord, scaling = 3, angle=15, type="n")
points(pl, "points", pch=16, col="red", cex = 0.7)
### identify(pl, "arrows", col="blue") would put labels in better positions
text(pl, "arrows", col="blue", pos=3)
text(pl, "centroids", col="blue", pos=1, cex = 1)
### Add species using xyz.convert function returned by ordiplot3d
sp <- scores(ord, choices=1:3, display="species", scaling=3)
text(pl$xyz.convert(sp), rownames(sp), cex=0.7, xpd=TRUE)
### Two ways of adding fitted variables to ordination plots
ord <- cca(dune)
ef <- envfit(ord ~ Moisture + A1, dune.env, choices = 1:3)
### 1. use argument 'envfit'
ordiplot3d(ord, envfit = ef)
### 2. use returned envfit.convert function for better user control
pl3 <- ordiplot3d(ord)
plot(pl3$envfit.convert(ef), at = pl3$origin)
### envfit.convert() also handles different 'choices' of axes
pl3 <- ordiplot3d(ord, choices = c(1,3,2))
plot(pl3$envfit.convert(ef), at = pl3$origin)