Weird (very large) Confidence Intervals when backTransforming occupancy (issue with the coefficient argument, I think)

[Luís Santiago]

Hello there!

I am trying to estimate mammal occupancy.

For one of the species I am covering, it turns out that I am not exactly getting an error report but, whilst the detection estimates look ok, the occupancy ones are looking rather odd (Estimate = 1; SE = 0.00251; 95% confidence intervals: LL = 0.008984463, UL = 1).

I suspect that things might have taken a wrong turn during the backtransformation step. The way I did it was based on the information provided by Ian Fiske and Richard Chandler in their package overview (2.3 in https://cran.r-project.org/.../unm.../vignettes/unmarked.pdf), but I don't know if I misunderstood the authors here or made a mistake elsewhere. In the authors' example, their coefficient argument (when using the backTransform in the fm2 model) is c(1, 0, 0), as they want to estimate detection and, in fm2, detection is modelled by two covariates.

I thought that the same would be true if my top model had occupancy being modelled by two covariates. However, my occupancy estimates seem to be odd.

I wonder where the wrong turn was (?) especially because, in species where the top model has only one covariate, or none, modelling occupancy/detection, the estimates for occupancy/detection come out alright. I tried to attach a few tables and a script regarding this issue, but it is not working, so please find them here:

https://wetransfer.com/downloads/c60c15c2e18b481c3b00d8611194af9f20220112193700/e0ff96c8a6c61f27c80c7984a214a26a20220112193743/845c68?fbclid=IwAR0YMXZ0E59U7Wtmn0VSkNeF79rJmjW8FByWj_rshvcPDB38JT6G5qxIaWE

. Let me know if you can't download them and I will dm them to you.

Thank you in advance!

Luís

[Luís Santiago]

Hi again,

In case this happens to anybody else, the way around this is running the code in a different R environment. I had a friend running my code in his computer and it worked fine and I ran it myself in my computer, just not in my "Occupancy environment". I don't know why it happens, but this is how you can dodge it, it is at least one of the ways.

Regards,

Luís