Multi-State Occupancy

[grandmaison]

I am working with a dataset of repeated detection/non-detection desert tortoise surveys. I've conducted an analysis where the detection of a live tortoise is treated the same as the detection of desert tortoise sign (i.e., scat, tracks, etc.). Now I'm interested in a multi-state approach with 3 states: (0)no detection, (1)live tortoise detected, or (2)sign but no tortoise detected.

I've run into a few questions after reading the multi-state occupancy literature. There is a subtle, but potentially important difference in the data I'm looking at relative to the work reported in the MacKenzie et al. 2009 spotted owl example. In their paper, psi is the probability that a site is occupied and R is the probability of successful reproduction at a site given that the site is occupied.

In the tortoise dataset, there can be no detection, a live tortoise detection, or sign detection (if both tortoise and sign, I've been categorizing it as 'live tortoise detection'). Neither state is necessarily predicated on the other. Certainly if sign is detected, you could assume that the site was used at some point. However, unlike MacKenzie et al. 2009 there is still uncertainty in the occupancy of a live tortoise even when sign in detected. Whereas if a breeding spotted owl pair is identified, we know for certain that the site is occupied by spotted owls.

The questions:
1. will the PRESENCE model account for uncertainty regarding tortoise occupancy even though sign was detected at a site? Is this 'delta'?
2. is it valid to collapse the 'state' to live tortoise when both live tortoise and sign are detected? or is it more appropriate to create a third 'state'?
2. am I over-thinking this thing and should I just state the assumption that if sign is detected, the plot is considered occupied even though the tortoise may have not been detected?

Thanks in advance for any comments you might have.

[dhewitt]

Along the lines of (3), I'd be worried about what your "sites" are and how long the season is. If you can be certain (as you said) that finding sign means the site is occupied, and you have no issues with closure (the intersection of site and season definitions), the Yes -- I think you are overthinking it. But the site and season issues might be big depending on the mobility of the critters and the layout of the sites.