Dear Colleagues,
I’m using PCord to do Indicator Species Analysis.  I have read the McCune 
and Grace book, but I still have two questions that I think I know the 
answers to, but want more opinions.  (Perhaps I missed the answers in the 
book).

1) Do I need to have equal number of sample units within each group?
For example, if I compared three habitats, and had 54 sample units from 
HabitatA, 35 from HabitatB, and 6 from HabitatC.  Then, even if a species 
has only one individual total represented, and it was collected from 
HabitatC, it can return a significant p-value from the Monte Carlo test.

2) What is the Monte Carlo test saying?
a) The Indicator Value given is correct (or at least higher... no lower)
b) This species is a good indicator species (at least statistically 
speaking).

It seems like it must not be (b), because I get significant Indicator Values 
that are very low (8.5 is the lowest, but many ranged from there to 25). 
[IVs can range from 0 to 100; 100 being no error: species is faithful and 
exclusive to particular habitat].

I see papers that have significant but low Indicator Values; some say these 
spp are good indicator species, others just let us look at the list w’ the 
IV and don’t really comment all that much on how faithful/exclusive those 
species are to a particular habitat.

I would appreciate your insights!
Thank you,
Shelly

Reply via email to