Data set suitable for occupancy modelling?

[richard]

Hi all,

I am working with single season occupancy models for point count data of ~10 birds using a data set that was initially not designed (but later adapted) for this kind of analysis
and was hoping that you could give me some recommendations as to whether or not the approach I am using is feasible.

My data set includes 630 sites with the following distribution of (re)visits:

Visits Sites
1-------416
2-------79
3-------56
4-------33
5-------20
6-------14
7-------2
8-------3
9-------2
10------2
11------1
12------1
14------1


1) I know that the modelling framework is robust towards "missing" observations, but given that 2/3 of the sampling sites were only visited
once I was wondering if you think this data set is suitable for analysis in PRESENCE/it's occupancy model framework?

2) We have chosen an exploratory approach in terms of covariable selection, resulting in a maximum of 4 site specific covariables on the p side and
25 on the psi side. For model selection we co-developed a stepwise AIC procedure in R employing the PRESENCE equivalent R package unmarked
for running the occupancy models. As I am getting some convergence issues for some of the birds both in R and when manually running them with PRESENCE
I was wondering if you think that the number of variables might be too high given the data set?

3) In terms of assessing model fit I am familiar with the test of the global model using parametric bootstrapping. I haven't really found anything
in terms of the validation of the model averaging results and was wondering what you are using for validation of the predictions from the averaged
models?



Thanks in advance for any suggestions you have in this matter,
Richard