To separate the effect of local or regional abundance of a tree
species from the effect of the tree species itself on the local rank
order (worth parameter), we fitted again a linear mixed effects
model with the worth parameter on each subplot per tree species
and year as a dependent variable; tree species, year and their
interaction as fixed effects; and stand and regional tree abundance
as covariates. To account for the nested observation, we
used a nested random factor of subplot, plot and region. The P
values adjusted for multiple comparisons were calculated again
using glht.