[R] NLME: Limitations of using identify to interact with scatterplots?

[Greg Distiller]

I have a quick question regarding the use of identify to interact with 
points on a scatterplot. My question is essentially: can identify be used 
when one is plotting model objects to generate diagnostic plots? 
Specifically I am using NLME.
For example, I am plotting the fitted values on the x axis vs a variable 
called log2game with the following code:

plot(D2C29.nlme, log2game ~ fitted(.), abline=c(0,1))

and then I have tried to use identify as follows:

identify(D2C29.nlme$fitted[,2],Data2$log2game,row.names(Data2))

(if I leave out the [,2] on the fitted attributes then I am told that x and 
y are not the same length and it appears that this is due to the fact that 
the fitted attribute has 2 columns.)

but I get an error message that "plot.new has not been called yet".

I am not sure if this is because I am doing something wrong or if identify 
simply cannot be used in this context.

Many thanks

Greg