[R-sig-eco] error strata in R

[Kingsford Jones]

Hi Nick,

Unfortunately if there is just one age per flow and there are only 3
flows, there is no replication to test for age effects.  Also w/ one
observation per Flow/Loc/Topo cell, the topo effect is confounded with
the error term.

Kingsford Jones

On Fri, Jan 22, 2010 at 7:52 AM, Nick Cutler <nick.cutler at ouce.ox.ac.uk> wrote:
> I'm working on long-term vegetation development on lava flows. I'm
> interested in the relationship between various ecological and environmental
> response variables (e.g. days of snow cover, soil temperature, foliar
> nutrient concentrations) and two fixed
> factors: microtopographic position ('Topo', with two levels: 'hillock' and
> 'hollow') and lava flow age ('Age', with three levels: '26_years',
> '161_years' and '848_years').
>
> Three lava flows of different ages were selected for the study. Sampling
> was carried out at three locations on each flow; each location comprised
> paired 'hillock' and 'hollow' plots (below).
>
> Flow            Location        Topo
> 26_years        1               Hillock
> 26_years        1               Hollow
> 26_years        2               Hillock
> 26_years        2               Hollow
> 26_years        3               Hillock
> 26_years        3               Hollow
> 161_years       4               Hillock
> 161_years       4               Hollow
> 161_years       5               Hillock
> 161_years       5               Hollow
> 161_years       6               Hillock
> 161_years       6               Hollow
> 848_years       7               Hillock
> 848_years       7               Hollow
> 848_years       8               Hillock
> 848_years       8               Hollow
> 848_years       9               Hillock
> 848_years       9               Hollow
>
> Generally, there was only one observation per cell. In this design, there
> are replicate observations of hillocks/hollows on each flow, but these
> observations are paired (a bit like a split plot design, or what Quinn &
> Keogh call a partly-nested ANOVA).
>
> I would like to analyze the effect of Age and Topo on the response variables
> (taking each variable individually). My initial thought was to use aov with
> Age and Topo as fixed effects, and Location (a blocking factor) as a random
> effect. But I am not sure how to specify the error term to account for the
> blocking. Would lme be more appropriate in this case?
>
> Can anyone help with the model specification?
>
> Nick
>
> --
> Dr Nick Cutler
> Postdoctoral Research Assistant
> School of Geography and the Environment
> Oxford University Centre for the Environment
> Dyson Perrins Building
> South Parks Road
> Oxford OX1 3QY
> UK
>
> Tel.: +44 1865 275 860
>
> _______________________________________________
> R-sig-ecology mailing list
> R-sig-ecology at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-ecology
>