[R-sig-eco] Residuals of observed values against linear reationship

[Sam]

Dear list,

I am having trouble formulating the syntax of my model.

I have two factors A + B these relate to quantities in global regions throughout the world ( 600). i have established their relationship using a GLS with a correlation structure accounting for spatial autocorrleation.

model <- gls(A~B, correlation = exponential)

what i want to know is in which regions A is higher than expected given the level of B - and the other way round. 

I know i need to use the residuals of observed A compared linear relationship between A and B this will show me  areas of lower than expected A (according to the B present in the area).

However i am unsure how to do this?

Any help would be greatly appreciated.