[R] species projected in a ordiplot

[Jari Oksanen]

santi8 <santi8 <at> unisi.it> writes:

> 
> Dear all,
> I'm performing a detrended correspondence analysis on vascular plant
> community data (296 species), and I have a question on the species scores
> projected in the ordination diagram. When I run a ordiplot all species are
> projected in the output graph, but I'd like to restrict the number of
> species plotted in the final graph. Some species are so rare in the data,
> that no relevant information is provided on their ecological preferences,
> and other species might not characterize well the analysis, then I'd like
> to select these species with highest fit state. 
> 
Dear Elisa Santi,

The question seems to concern a contributed package (vegan), but not many people
see that from your message: please give these details when posting. Moreover,
R-sig-ecology mailing list would probably give faster and more numerous
responses. Here's one response, though. 

You do not need to use generic ordiplot() to plot decorana() results: the
function has its own plot() method. The plot() method (just like ordiplot())
gives a quick standard plot. If you want more control, you should first generate
an empty plot with argument type = "n", and then use text() or points() to add
items you need. The text() and points() functions for the decorana plot() have
argument 'select' that can be used to select the points you want to have. The
select can be a logical vector that is TRUE for selected items or a numeric
vector of indices of selected items. The following example displays some most
abundant species in the dune data set:

ord <- decorana(dune)
plot(ord, type="n")
text(ord, display="species", select = colSums(dune) > 20)

Ciao, Jari Oksanen