vegan (version 2.6-6.1)

ordiresids: Plots of Residuals and Fitted Values for Constrained Ordination

Description

The function provides plot.lm style diagnostic plots for the results of constrained ordination from cca, rda, dbrda and capscale. Normally you do not need these plots, because ordination is descriptive and does not make assumptions on the distribution of the residuals. However, if you permute residuals in significance tests (anova.cca), you may be interested in inspecting that the residuals really are exchangeable and independent of fitted values.

Usage

ordiresids(x, kind = c("residuals", "scale", "qqmath"),
   residuals = "working", type = c("p", "smooth", "g"),
   formula, ...)

Value

The function returns a Lattice object that can displayed as plot.

Arguments

x

Ordination result from cca, rda, dbrda, capscale.

kind

The type of plot: "residuals" plot residuals against fitted values, "scale" the square root of absolute residuals against fitted values, and "qqmath" the residuals against expected distribution (defaults qnorm), unless defined differently in the formula argument.

residuals

The kind of residuals and fitted values, with alternatives "working", "response", "standardized" and "studentized" (see Details).

type

The type of plot. The argument is passed on to lattice functions.

formula

Formula to override the default plot. The formula can contain items Fitted, Residuals, Species and Sites (provided that names of species and sites are available in the ordination result).

...

Other arguments passed to lattice functions.

Author

Jari Oksanen

Details

The default plots are similar as in plot.lm, but they use Lattice functions xyplot and qqmath. The alternatives have default formulae but these can be replaced by the user. The elements available in formula or in the groups argument are Fitted, Residuals, Species and Sites.

With residuals = "response" and residuals = "working" the fitted values and residuals are found with functions fitted.cca and residuals.cca. With residuals = "standardized" the residuals are found with rstandard.cca, and with residuals = "studentized" they are found with rstudent.cca, and in both cases the fitted values are standardized with sigma.cca.

See Also

plot.lm, fitted.cca, residuals.cca, rstandard.cca, rstudent.cca, sigma.cca, Lattice, xyplot, qqmath.

Examples

data(varespec)
data(varechem)
mod <- cca(varespec ~ Al + P + K, varechem)
ordiresids(mod)
ordiresids(mod, formula = Residuals ~ Fitted | Species, residuals="standard",
   cex = 0.5)