Interpreting site estimates for psi

[mtbikernate]

I have used the detection histories of other species as covariates in my analysis process (camera trap data). When doing so to estimate p, the output makes sense. It's messy, but I'm only given estimates at a site when surveys are active at that site.

When using one of these covariates to estimate psi, the results are odd to me. Everything else about the output makes sense. I get no errors. c-hat is 1.0278. DeltaAIC is <2 for this particular model. But I get the following output for the individual site estimates of psi.

Individual Site estimates of <psi>
Site estimate Std.err 95% conf. interval
psi 1 2 . . . - .
psi 2 3 . . . - .
psi 3 5 . . . - .
psi 4 6 . . . - .
psi 5 7 . . . - .
psi 6 9 . . . - .
psi 7 148 . . . - .
psi 8 1 : 0.0000 0.0000 0.0000 - 0.0000
psi 9 148 . . . - .

Part of the problem might be high occupancy. If I look at the derived site estimates, the dots in the above output correspond to a derived occupancy of 1, and 0 for site number 8.

Site psi-cond Std.err 95% conf. interval
1 2 1.0000 0.0000 1.0000 - 1.0000
2 3 1.0000 0.0000 1.0000 - 1.0000
3 5 1.0000 0.0000 1.0000 - 1.0000
4 6 1.0000 0.0000 1.0000 - 1.0000
5 7 1.0000 0.0000 1.0000 - 1.0000
6 9 1.0000 0.0000 1.0000 - 1.0000
7 148 1.0000 0.0000 1.0000 - 1.0000
8 1 0.0000 0.0000 0.0000 - 0.0000
9 148 1.0000 0.0000 1.0000 - 1.0000

Naive occupancy is 0.7778. Other plausible models are providing site estimates in the range of 0.1-0.89. Is the output telling me that the site estimates are 1 when it is reporting a "." or is this indicative of an error of some sort and I should exclude this model from further analysis?

[darryl]

Do any of the covariates you're placing on psi have missing values?

[mtbikernate]

yes, there are missing values.

the matrices for the covariates I'm using in these cases match the detection histories for the species I'm looking at (with missing values on the same survey occasions). but there are missing values for all of the sites, not just the ones reported with dots. if it was a missing value problem, I could see the site parameters giving me odd outputs for every site with missing values (all of them).

this is giving me dots at sites where the species was observed and 0 where it was not.

should I not be using these variables as covariates for psi?

[darryl]

So, you're using the detection history for another species, entered as a sampling occasion covariate, as a covariate for psi?

If so, then PRESENCE will only be using the first column of that covariate for psi, not all of them as really you should be entering covariate for psi as a site-specific covariate. Check out some of the information in the help menu.

Do the sites with dots reported line up with those sites that have missing values in the first column of the covariate?

[mtbikernate]

ohhh, I see. I see.

Yes, the dots line up with the missing values in the first column. that makes sense.

Forgive me for being a little slow.

We met at the Texas Chapter of TWS meeting in 09...ate lunch together at Bob Evans, IIRC. a week or two after that, I got a cancer diagnosis and spent some time in a coma. coming back from that has been tough.

[darryl]

Sorry to hear that, but glad to hear you're on the mend.