[R-sig-eco] life forms spectrum analysis

[Rich Shepard]

On Mon, 16 Mar 2015, Ludovico Frate wrote:

> No, I'm interested in the whole plot composition! I would like to
> understand if there are differences in terms of life forms composition
> beteween two groups of plots (each group has 70 plots). A simple t-test or
> other non parametric tests do not work fine with count data, as in my
> case. In addittion some life forms are uncommon so I have a lot of zero
> count in my data!

Ludovico,

   Then you definitely should look at CoDA because, as you write above, you
are looking to see if there are differences betwee the compositions of the
two groups, and there will be compositional differences within each group
that need to be distinguished from the between-group compositions.

   ven den Boogaart and Tolosana-Delgado's book, "Analyzing Compositional
Data with R" is highly recommended. There are also abundant references on
the Web, such as Pawlowsky-Glahn, Egozcue, and Tolosana-Delgado. 2007.
Lecture Notes on Compositional Data Analysis.

HTH,

Rich