News

Since R version 4, data imported using read.table() or read.csv() functions do NOT have character columns converted to factor variables.

This means that examples where a factor variable is assumed will not work as expected, and you will need to alter character variables to factor.

You can do this individually for each variable required for example:

names(pd)
[1] "Site" "Species"
pd$Site <- as.factor(pd$Site)

You can also convert all character variables to factor using the rapply() function like so:

pd <- rapply(pd, f = as.factor, classes = "character", how = "replace")

In this case any columns that were originally character would be converted to a factor.

polar.ord()

This is a new command that conducts polar ordination (also called Bray-Curtis ordination). The book shows polar ordination and presents a method to carry it out using Excel (Chapter 14). This is labour intensive but instructive for students, as it shows some of the basic principles behind ordination in general. However, I thought it would be useful to have an R command to carry out the procedure.

The command works out polar ordination site scores and works out axes as follows: Axis 1 is the combination of sites that gives the maximum separation (not necessarily the maximum variance explained). Axis 2 is the combination of sites that maximises the orthogonality to the 1st axis.

There are print, plot and summary methods and an identify method for use with plots.

polar.ord(diss)

print.polar.ord(ord, digits = getOption("digits"))

summary.polar.ord(ord, k = 2, cor = TRUE, var.exp = TRUE, digits = getOption("digits"))

plot.polar.ord(ord, x.axis = 1, y.axis = 2, type = c("t", "p", "n"), ...)

identify.plot.polar.ord(pord, labels = pord$labels, ...)

See examples in the Custom R Commands page.

You can get a copy of the .R file here. Click the link to open the text. Right-click and “Save As” to download the file.

The routines are in the main data file. However, if you get the Polar Ordination.R file separately, it opens in any text editor. To install the commands in R, use something like:

source(file.choose())

Then select the file. In Linux you’ll need to specify the filename explicitly.

plot_H()

This formerly computed bootstrap confidence intervals only (for “shannon” or “simpson”). Now you can specify "shannon", "simpson", or "invsimpson", which use H_CI() and methods using variance estimates. The command calculates the appropriate diversity index and CI, then plots the result.

NOTE: returns an error if you specify boot = TRUE and index = "invsimpson".

Iapp()

This compliments the functions Happ() and Dapp(). It calculates bootstrap CI for Inverse Simpson’s index ("invsimpson"). Designed for use with apply() and others as an “internal” command.

aic()

This is a wrapper for add1() that allows display of potential variables in order of AIC. You can specify order (default: decreasing = FALSE) and number of “results” to display. Specify n = "all" to show all results (the default), or integer to display n results. This is useful when your pool of potential variables (the scope parameter) is large.

Some changes in recent versions of the vegan R-package prompted a few alterations.

The vegan team removed Std Err calculations from fisherfit(). There are sensible reasons for doing so as CI are somewhat “dodgy”. However, to get the same results as the exercises in the book I restored the behaviours. If you want to get the same output as the book then use fisher_alpha() and fisher_fit() in place of fisher.alpha() and fisherfit() commands.

fisher_alpha()

Renamed from fisher.aplha() and restores calculation of se and allows profiling.

fisher_fit()

Renamed from fisherfit(), which allows se calculations and profiling using the following commands restored from vegan 1.17.12.

confint.fisherfit()

plot.profile.fisherfit()

profile.fisherfit()