[R-sig-eco] Detecting spatial gradient whilst accounting for spatial autocorrelation

[Peter Francis]

Dear List

I have a data set which shows high levels of spatial autocorrelation, the sampled plant communities are on a global scale and i am looking for correlates of different diversity measures (PD,FD).

I have controlled for the spatial autocorrelation using a GLS model where the latitude and longitude of each ecoregion was included as a smooth factor in all-statistical models using the corExp function. 

i.e

exponential <-corExp(form = ~ Longitude + Latitude)

PD_global <- gls(PD~Elevation+Temperature+Species_Richness, correlation = exponential)

Now i want to ask the question; how do these diversity measure vary with latitude and longitude i.e is their a spatial gradient to PD or FD, i don't think i can put in longitude or latitude as a predictor if i am using it as a random variable through the corExp function 

i.e 

PD_global <- gls(PD~Elevation+Temperature+Species_Richness, ALtitude + Longitude, correlation = exponential)


But i imagine it is something that is often asked?

Any help would be greatly appreciated.

Thanks in advance

Peter