[R-sig-ME] Mixed models and time series

[Roberto Mannu]

Dear,

First of all I want to apologize for my not perfect English, I'll try to well explain my issues too. At the moment I'm trying to
analize a dataset of three-years insect counts. Insects were counted
monthly on 4 sampling trees (causally selected) in 12 different
locations. Therefore, I firstly obtained 12 similar dynamics for each
location. My first question is: "Can I evaluate differences in
population among location?". I thought to fit a linear mixed model after
data log-transformation in which "location", "time" (evaluated as the
time from beginning to end of sampling) and its relationship were the
fixed factors and sampling trees (1 to 4 within each location at each
sampling date) were the randoms factors. Can this being considered right?

I'm afraid that there may be temporal autocorrelation between data cause
of regular dynamics within year.

I read different work regarding GLMMs, but I did not understand how to apply
them in R.

Thanks in advance for your reply.

Best regards


Roberto


-- 
Roberto Mannu, PhD
Istituto per lo Studio degli Ecosistemi
Consiglio Nazionale delle Ricerche
Traversa La Crucca 3
07100 Sassari

tel: +390792841410
cell: +393401612546
  
email: r.mannu at ise.cnr.it