[R-sig-eco] split-split-plot design and adonis

[Steve Brewer]

Arnaud,

Have you already tried to run the analysis using the Site_Hydro
interaction as the strata effect? If so, can you show us the anova table?

One thing I'm wondering about is how adonis is dealing with the apparent
nesting of sites within sediment types. From your description, site does
not appear to me to be equivalent to a "block" at the level of sediment
type (although it does appear so at the level of hydrology).

I've not been successful at doing either a split plot or a repeated
measures using Adonis. I don't know whether it is because I have set up
the data frame containing the factors incorrectly or whether the arguments
or models in adonis I'm using are incorrect, but the resulting anovas I
get always show the residual error term being used to test all the
effects, which is incorrect and not what I want.

Best,
Steve

 
J. Stephen Brewer 
Professor 
Department of Biology
PO Box 1848
 University of Mississippi
University, Mississippi 38677-1848
 Brewer web page - http://home.olemiss.edu/~jbrewer/
FAX - 662-915-5144
Phone - 662-915-1077




On 3/1/12 3:31 AM, "Arnaud Foulquier" <arnaud.foulquier at gmail.com> wrote:

>Dear colleagues,
>
>I¹m working on a data set in order to evaluate the impact of drying on
>sediment microbial community structure. The objective is to determine if
>the
>impact of drying varies among sediment types and/or depth within the
>sediment.
>
>The experimental design is as follows: The first factor *Sediment*
>corresponds to three types of sediment (coded Sed1, Sed2, Sed3). For each
>type of Sediment, sampling was carried out on three sites (3 sites for
>Sed1,
>3 sites for Sed2, 3 sites for Sed3). *Site* is coded : Site1, Site2, ...,
>Site9. The next factor is *Hydrology*: within each site, sampling is
>carried
>out in a dry plot and in a wet plot (coded Dry /Wet). Within each of the
>previous plot, sampling is carried out at two *Depths* (D1, D2) in
>triplicate.
>
>There are a total of n = 108 samples = 3 Sediment * 3 Sites * 2 Hydrology
>*
>2 Depths * 3 Replicates.
>
>I use the adonis function in the following way :
>
>Sediment<-as.factor(rep(c("Sed1","Sed2","Sed3"),each=36))
>Site<-as.factor(rep(c("Site1","Site2","Site3","Site4","Site5","Site6","Sit
>e7","Site8","Site9"),each=12))
>Hydrology<-as.factor(rep(rep(c("Dry","Wet"),each=6),9))
>Depth<-as.factor(rep(rep(c("D1","D2"),each=3),18))
>
>adonis(community~Sediment*Hydrology*Depth,perm=999,strata=Site)
>
>I need to specify strata=Site to account for the stratified structure of
>the
>design.
>However, I think it is incomplete and that I also should specify that
>Hydrology is nested in Site (I use the lme function in the univariate case
>as follows :
>http://stats.stackexchange.com/questions/19818/split-split-plot-design-and
>-lme)
>
>I tried the following :
>
>Site_Hydro<-interaction(Site,Hydro)
>adonis(community~Sediment*Hydrology*Depth,perm=999,strata=Site_Hydro)
>
>Is it the correct way to account for the stratified structure of my data?
>
>Any help on this question would be much appreciated.
>
>Arnaud
>
>
>
>
>
>--
>View this message in context:
>http://r-sig-ecology.471788.n2.nabble.com/split-split-plot-design-and-adon
>is-tp7332029p7332029.html
>Sent from the r-sig-ecology mailing list archive at Nabble.com.
>
>_______________________________________________
>R-sig-ecology mailing list
>R-sig-ecology at r-project.org
>https://stat.ethz.ch/mailman/listinfo/r-sig-ecology