closure assumption and spatial replicates

[giancarlo]

Bill Kendall and Gary White recently came out with an interesting paper (2009, J. Appl. Ecol. 46:1182-1186) showing how occupancy estimates can be inflated when the closure assumption was violated across spatial replicates.

I was at the NCTC occupancy workshop last month and Jim Nichols discussed the use of the spatial dependency settings in the single-season model. The application seemed specific to one study of animals on trails, but I'm wondering about its broader application.

From reading this thread: http://www.phidot.org/forum/viewtopic.php?t=1162 and others, it seems this custom model can act as a "work around" to the types of issues encountered when only an (unknown) portion of spatial replicates within a site are "available" for occupancy. Clearly, getting temporal replicates of spatial replicates (as suggested by Kendall and White) would be ideal, but what to do with data containing only spatial replicates?

I've seen a few recent studies that used spatial replicates without discussion of these potential closure violations. Often survey covariates are added to account for the non-detections/'true absences' among some replicates, but this doesn't seem to get at the issue raised by Kendall and White.

Any ideas? And, if the custom spatial dependency model is the right tool for the job, anyone know if Jim Hines is planning to add it to the multi-season models?

Best,

Giancarlo