R help - May 2009 - glht problem

I am struggling with a simple repeated-measure model:

fit<-lme(trait~year * A, random = ~1|subj/year)

A being a factor with three levels. I got have the following results
for  anova(fit):

            numDF denDF   F-value p-value
(Intercept)     1   126 2471.4720  <.0001
year           20    60   10.4126  <.0001
A                 2   126   23.0721  <.0001
year:A        40   126    1.6499  0.0193

Now I try to use glht for A, but fail:

Linear Hypotheses:
                      Estimate Std. Error z value p value
A2 - A1 == 0      0.25       1.10   0.227   0.972
A3 - A1 == 0     1.00        1.10   0.909   0.634
A3 - A2 == 0      0.75       1.10   0.682   0.774
(Adjusted p values reported -- single-step method)

Warning message:
In mcp2matrix(model, linfct = linfct) :
  covariate interactions found -- default contrast might be inappropriate


What can be going on with this?

many thanks in advance,
Wolf