[R-sig-eco] anosim, mrpp and adonis

[Jari Oksanen]

Roman,

On 17/12/2009 18:29, "Roman Luštrik" <roman.lustrik at gmail.com> wrote:
> 
> I'm trying to see if samples within and between locations differ (in case
> they don't, I will have an argument to join subsamples into one). I have
> tried ANOSIM and mrpp.
> 
> anosim.result <- with(environmental.sub, anosim(community.sub.dist,
> station)) #station is used as "strata"
> mrpp.result <- with(environmental.sub, mrpp(community.sub.dist,
> station))#ditto

> 
> My question would be, in what way can I see between and within differences
> in mrpp (can I plot this?)

Not directly in mrpp(), but you can use meandist() which is documented
together with the mrpp() and has a plot method (in vegan 1.15-4 at CRAN and
vegan devel). The meandist() function is just another formulation of mrpp
(or vice versa). 

> and how would a similar method be done in
> adonis() (both ANOSIM and mrpp help files link to adonis as a more
> appropriate method)? I have tried as displayed below, but clearly this is
> not the solution (or am I missing something?). Can anyone advise on how to
> improve it to match what I'm trying to achieve with ANOSIM/mrpp?
> 
> 1> adonis.result <- adonis(community.sub.dist ~., data=environmental.sub,
> strata=environmental.sub$station, permutations=999)
> 1> adonis.result
>
If you want to fit a similar model as with mrpp and anosim above, you should
use

adonis(community.sub.dist ~ station, data = environmental.sub)

Then you cannot have 'strata = station' which would stratify all
permutations so that in no case would the 'station' change in fitted models
and all permutations would have identical statistics (within numerical
accuracy). 

adonis() does not have plot method, but  you can use betadisper for plots.
 
Cheers, Jari Oksanen