spatial correlation / thta0i NOT equal rest of thta0's

[shannonbarbermeyer]

Hi - I'm running the single season spatial correlation model in Presence using data that was collected in spatially sequential segments along trails. Each 'site' included 40km of continuous walking - 40 connected sequential segments of 1km (each considered a survey or spatial replicate). When I took the Presence training there wasn't a box under "Custom w/ spatial correlation" to check for "initial thta0 not = rest of thta0s". What is the interpretation of this in terms of what it would mean "on the ground"? When would you want to check this box? The default model without checking the box already has theta (species present on current segment and not on previous segment) and theta prime (species present on current segment and on previous segment), right? So would checking the box mean that the probability of a 1st zero on the 40km walks then being followed by either a 1 or 0 would not equal the remaining 0's probabilities of those same events? What would be a possible reason this kind of correlation would occur versus the default one? Also, why does Presence still only estimate 4 parameters (same as the default spatial correlation) rather than an extra one if these theta0's are not constrained to be the same? THANK YOU for your help!

[jhines]

Shannon,

In the spatial correlation model, you have no information about the occupancy or status of the site before the first survey in the segment. The original thinking was to treat the first survey the same as surveys where the previous survey was unoccupied. So, theta0 for the first survey would be the same as theta0 for any other survey.

After some thought, it made more sense for the study we developed the method for, to treat the first survey differently, since the segment before the survey might be occupied. We decided to use a weighted value between theta (previous survey unoccupied), and theta' (previous survey occupied). The weighting factor would be the prob. of occupancy, so no extra parameter is needed.

If your study design is such that the first survey in a segment is always at an edge, where the segment before it is always unoccupied, then you would leave the checkbox option unchecked (theta0 equal...). If the first survey in a segment is placed randomly in space, then you would check this option off.

Cheers,

Jim

[shannonbarbermeyer]

Thank you Jim! That explains it very well. Super helpful.

Shannon