[R-sig-eco] Fisher method and Monte Carlo

[Mario José Marques]

Hi all,

my question is more statistic than r-ecology.

I am running a protocol to fit niche apportionment species abundance
distribution (SAD) proposed by Muillot et al (2003). This protocol is a
modification of Bersier and Sugihara (1997) and Cassey and King (2001).

Briefly, this method use Fisher Method (combining independent p-values)
to compare with a random value. A specific SAD stochastic model run X
randomizations of SAD, then observed abundance of each specie is
compared with X random value to this species. Each specie then will have
a 'p-value' and the overall p-value is computed with Fisher Method: p =
-2 * sum(ln(pi)) (i is each specie).

For each specie, p-value is: p(i)=  2 min( (randomization(i) >
observed)/X, (randomization(i) < observed)/X )

The proposed protocol don't include observed value in randomization and
logical test is larger than or lesser than, and not larger or iqual
than, like proposed to Manly (Bootstrap, randomization and Monte Carlo).
This result in some p-values = 0 (for some species), and when use Fisher
Method I have some -Inf ( ln(0) = -Inf ).

My doubt is about this difference, how this different computation (the
protocol and the Manly) can modify statistic used (Monte Carlo)?

I read that is possible substitute 0 to 0.00001, and that have weight
approach to Fisher Method, but this is used when Fisher Method is to
compute statistic to X². The protocol only use the value (a measure of
how far my observation is of simulated values) to compare with random
values using Monte Carlo approach.

Excuse me if this question is quite off topic.

Best regards,

Refs.

Bersier L-F, Sugihara G (1997) Species abundance patterns: the problem
of testing stochastic models. J Anim Ecol 66:769–774. doi: 10.2307/5927

Cassey P, King RAR (2001) The problem of testing the goodness‐of‐fit of
stochastic resource apportionment models. Environmetrics 12:691–698.
doi: 10.1002/env.493

Mouillot D, George‐Nascimento M, Poulin R (2003) How parasites divide
resources: a test of the niche apportionment hypothesis. J Anim Ecol
72:757–764. doi: 10.1046/j.1365-2656.2003.00749.x


Mario José Marques
...................................................
Doctoral student in Ecology
Institute of Biology, Dept. Plant Biology
State University of Campinas - UNICAMP
Campinas, São Paulo, Brazil