Trouble with random factors glmm

#1
Hi.
I am doing a behavioural ecology study on juvinile brown trout and want to perform a glmm (glmer, package lme4) to analyse the effect of activity, size and treatment on survival. The variables i am using and are: response-survival (dead or alive), fixed effects- size (continuous), activity level (continuous) and treatment (fully fed or restricted diet). However, when I was keeping the fish at laboratory I grouped them in twenty compartments 10 by 10 with only one treatment in each compartment. This obviously make me worry about pseudoreplication and I understand thats wy I should add a random factor. I don't really get how to write this random factor in the model and if it should be nested with(in?) treatment (or individual).

Right now my models looks like this:

glmer(survival~size+activity+treatment+(1|compartment), family=binomial)

Does this seem right to you?

I actually have this problem when doing a linear mixed effect model (lme) analysing the effect of treatment (as above) over time (before and after treatment) on activity as well.
That model looks like this right now:

lme(activity~treatment*time, random=~(1|compartment/time))

I would be very thankful for some help.

B//Per S