[ORDNEWS:1833] Additional partial environmental variables in PRC

[Roy Sanderson]

Dear Ordnews

I am undertaking a Principal Response Curve analysis and wish to partial out the effects of additional environmental variables, in addition to the usualTreatment (control and teatment levels) and Time variables. A conventional PRC is a type of RDA set to:

rda(response ~ treatment:time + Condition(time))

The first level of the treatment is assumed to be the 'Control' and set to zero in the PRC plot. I wish to include additional environmental variables (in this case continuous ones), but partial out their effects to understand more clearly the treatment effects over time. Assuming that 'envvar' is the new environmental variable, is the best structure for this:

rda(response ~ treatment:time + Condition(envvar) + Condition(time))

or should envvar be additionally included in a 3-way interaction term?

The R code underlying the prc command in vegan is fairly clear, and so it should in theory be relatively straightforward to make any changes. However, I was also unsure how to partial out any effects that an additional environmental variable would have at the 'Control' treatment level. In the prc R code the Control level is deleted from the model design matrix prior to calling the rda command itself?

Many thanks for your help
Roy



Roy Sanderson
School of Biology
Ridley Building
Newcastle University
Newcastle upon Tyne NE1 7RU
roy.sanderson-pyHcBy/D9uhaa/***@public.gmane.org
Tel: [phone]