ggvegan is a package for the R statistical software and environment. It aims to implement ggplot-based versions of the plots produced by the vegan package. Initially, ggvegan will provide fortify and autoplot methods for objects created in vegan, with the aim of providing full replacement plots via autoplot. The fortify methods allow the data contained within objects created by vegan to be converted into a format suitable for use with ggplot directly.
ggvegan is released under the GNU General Public Licence, version 2.
ggvegan uses the roxygen2 system to document package functions alongside the code itself.
ggvegan is very much alpha code at the moment. Comments and feedback on the approach taken are welcome, as are code contributions. See Design decisions below for two important areas for consideration
The autoplot concept is somewhat poorly defined at the moment --- at least in public. I have taken it to mean that a full ggplot object is returned, which can then be augmented with additional layers and changes to the scales etc. This means that the aesthetics for the scores are hard-coded in the autoplot methods. If you want greater control over these aesthetics, use fortify to return the scores in a suitable format and build the plot up yourself. I hope to include at least one example of this, where applicable, in the help pages for each autoplot method.
fortify methods are supposed to return a data frame but this is not necessarily the most convenient representation for vegan's ordination objects where several data frames representing the various sets of ordination scores would be more natural. Currently, ggvegan follows the existing fortify convention of returning a single data frame so returns the ordination scores in long format with variables Score indicating the type of score and Label the label/rowname for each score.
From version 0.0-9, I changed the representation of fortified ordination objects. The first two columns will now be Score and Label. The remaining columns will be the requested ordination dimensions, named as per the scores method from vegan. For example, a PCA will have columns named 'PC1', 'PC2', etc. How many and their numbering depending on the axes argument; the default is 1:6. Consequently, the 'dimLables' attribute is no longer necessary.
A further design decision is that ggvegan fortify methods for ordination objects will return all possible sets of scores and the set returned can not be chosen by the user. Instead, the sets of scores to be plotted should be chosen at the autoplot stage.
The components returned for more specialised objects will typically vary as needed for a sensible, tidy data representation. Such fortify() methods will return suitable components. For example, fortify.prc() returns components Time, Treatment, and Response corresponding to the two-way factors defining the experiment and the regression coefficients on RDA axis 1 respectively.
The following autoplot methods are currently available
autoplot.cca--- for objects of classes"cca"and"capscale"autoplot.rda--- for objects of class"rda"autoplot.metaMDS--- for objects of class"metaMDS"autoplot.prc--- for objects of class"prc"autoplot.decorana--- for objects of class"decorana"(AKA DCA)autoplot.prestonfit--- for objects of class"prestonfit"autoplot.fisherfit--- for objects of class"fisherfit"
The following fortify method are currently available
fortify.cca--- for objects of classes"cca","rda", and"capscale"fortify.metaMDS--- for objects of class"metaMDS"fortify.prc--- for objects of class"prc"fortify.decorana--- for objects of class"decorana"(AKA DCA)autoplot.prestonfit--- for objects of class"prestonfit"autoplot.fisherfit--- for objects of class"fisherfit"
No binary packages are currently available via CRAN for ggvegan. If you have the correct development tools you can compile the package yourself after downloading the source code from github.
You can install ggvegan directly from GitHub using functions that the remotes package provides. To do this, install remotes from CRAN via
install.packages("remotes")
then run
remotes::install_github("gavinsimpson/ggvegan")
If that doesn't work or you prefer to install from binaries, the R Universe service run by rOpenSci now provides binaries. Instruction on how to install ggvegan from that repository are:
# Enable repository from gavinsimpson
options(repos = c(
gavinsimpson = 'https://gavinsimpson.r-universe.dev',
CRAN = 'https://cloud.r-project.org'))
# Download and install ggvegan in R
install.packages('ggvegan')