life forms spectrum analysis

I would like to analyse the differences  in life form (Raunki??)
composition between two populations. The two populations (plant
communities) were sampled randomly and the presence of plant species were
recorded. So for each plot I have the frequency of each life form. First,
I thought to a chi square or binomial test. However, I am worried about
the applicability because both tests require that the frequency of life
forms observed vs expected is grater than 0% (i.e., when a given life form
is present in all the plot of both populations - observed = 100% expected
= 0%) and less than 100% (better between 20 and 80%).

Pop1, Pop2
present, 70, 56
absent, 0, 0

Do you have any ideas about analysing differences in life forms?

Ludovico,

   Let me paraphrase what you wrote to see if I understand. You have a number
of plots each populated with a variety of plant species, and you are
interested in only two of those species. You ask if there are meaningful
differences in the relative abundances of each of your two species of
interest in the plots.

   If this is correct (or close to correct), I suggest that you look at these
as compositions (the packages compositions, robCompositions, and
zCompositions each offer useful functions). The number (or percent ground
cover) of the species other than your two are immaterial. Assuming that the
relative proportions of each of your two species are the response
(dependent) variables, and you have potential explanatory (independent,
predictor) variables, then you can perform regression analyses on the
compositions.

Hope this helps,

Rich

Hi Ludovico,

You can compare the proportion of each life form using logistc regression, 
an example with simulated data for 20 plots:

set.seed(100)
n<-20
population<-sample(c("A","B"),n,replace=T)
lifeform1count<-round(runif(n,0,100),0)
otherlifeformscount<-round(runif(n,0,100),0)
f<-glm(cbind(lifeform1count,otherlifeformscount)~population,family=binomial)
summary(f)#significant difference between populations p=0.01137
plogis(coef(f)[1])#mean proportion of life form 1 in population A
plogis(coef(f)[1]+coef(f)[2]) #mean proportion of life form 1 in population 
B

You can run one model for each life form, but maybe multinomial regression 
is more correct (never tried)

Aitor

-----Mensaje original----- 
From: Ludovico Frate
Sent: Monday, March 16, 2015 6:07 PM
To: r-sig-ecology at r-project.org
Subject: [R-sig-eco] life forms spectrum analysis

Dear List,
I would like to analyse the differences  in life form (Raunki?r) composition 
between two populations. The two populations (plant communities) were 
sampled randomly and the presence of plant species were recorded. So for 
each plot I have the frequency of each life form. First, I  thought to a chi 
square or binomial test. However, I am worried about the applicability 
because both tests require that the frequency of life forms observed vs 
expected is grater than 0% (i.e., when a given life form is present in all 
the plot of both populations - observed = 100% expected = 0%) and less than 
100% (better between 20 and 80%).
Pop1, Pop2
present, 70, 56
absent, 0, 0

I have tried also a simple Mann whitney test but it doesn't work well for 
count data. In addition for uncommon life forms I have a lot of zero counts 
in my data!


Do you have any ideas about analysing differences in life forms?
Best,
Ludovico




Dott. For. Ludovico
Frate, Ph.D.
University of Molise - Italy
Environmetrics Lab
http://www.distat.unimol.it/STAT/environmetrica/organico/collaboratori/ludovico-frate-1
Department of Biosciences and Territory - DiBT
Universit? del Molise.
Contrada Fonte
Lappone,
86090 -  Pesche (IS)
ITALIA.
Cel: ++39
3333767557
Fax: ++39 (0874) 404123
E-mail ludovico.frate at unimol.it
ludovicofrate at hotmail.it
https://www.researchgate.net/profile/Ludovico_Frate








_______________________________________________
R-sig-ecology mailing list
R-sig-ecology at r-project.org
https://stat.ethz.ch/mailman/listinfo/r-sig-ecology