[R-sig-ME] Temporal and spatial correlation at the same time?

[Jaime Otero Villar]

Dear all,

I'm working with a set of data that comprise long time series of  
abundance of fish in different rivers. I'm modelling this abundance  
using covariates, random intercept and slope and serial correlation at  
the river level as follows:


model.lme1<-lme(abundance~temp+runoff,
+ 							random=~temp|River,
+ 							correlation=corAR1(form=~1|River),
+ 							data=myData)


This model assumes that there is no correlation of residuals for  
different time series, and I wonder if it would be possible to handle  
temporal and spatial correlation at the same time in nlme. Or, could  
anybody give some hint on how to take into account variability among  
rivers as a function of distance between them?


Thanks in advance for helping.


Jaime.