Predicting derived occupancy

posts related to the RPresence library, which may not be of general interest to users of 'classic' PRESENCE.

Predicting derived occupancy

Postby koustubhsharma » Sun Aug 16, 2020 11:03 pm

Dear all,
This might be a naive question, but is there a way to predict derived occupancy in a multi-season model for un-sampled units using RPresence?

For single season models, I typically use: predict(top.model, newdata=Data1, param="psi"), but am unable to find a way to predict derived psi from multi-season models. I believe one round-about solution would be that of including un-sampled units with no data in the pao along with covariates before the analysis. However will be grateful if there is an easier way to get this done.

Thanks much,

Posts: 2
Joined: Sun Jul 05, 2015 11:15 pm

Re: Predicting derived occupancy

Postby koustubhsharma » Mon Aug 17, 2020 12:14 am

Oh well, my apologies for posting a premature question without looking deeply. I guess I found the correct way (hopefully) to estimate the derived occupancy for subsequent seasons.

I used predict to estimate colonization and extinction probabilities for the unsampled units, and then used:
psi.1*(1-epsilon)+(1-psi.1)*gamma to predict occupancy for the subsequent seasons. While this works and provides the estimates of psi correctly, I may still need assistance with estimating errors and confidence intervals of the derived parameters!

Thanks once again, and please advise if I've got this wrong.


Posts: 2
Joined: Sun Jul 05, 2015 11:15 pm

Re: Predicting derived occupancy

Postby B.K. Sandercock » Mon Aug 17, 2020 2:48 am


You can obtain estimates of the variance and confidence intervals for your derived parameters with the delta method as an analytical method or with bootstrapping as a numerical approach. Both methods are described in detail along with sample R code in Appendix B of the Mark manual. Larkin's paper below is also a helpful introduction to the delta method:
Powell, L.A. 2007. Approximating variance of demographic parameters using the delta method: a reference for avian biologists. Condor 109:949-954.

Good luck, Brett.
B.K. Sandercock
Posts: 48
Joined: Mon Jun 02, 2003 4:18 pm
Location: Norwegian Institute of Nature Research

Re: Predicting derived occupancy

Postby darryl » Mon Aug 17, 2020 4:54 am

Hi Koustubh
To get the SE for the derived psi estimates you will need the full variance-covariance matrix for the initial psi, seasonal colonisation and extinction probabilities. There will be code within the depths of the package that you could explore how the calculations are done (including the delta method) and create your own function to replicate.

Alternatively, take the approach you outlined at the start and add in some dummy unsurveyed units to your input data so RPresence gives you the values as part of the occMod object. That might be fastest way at the moment.

Posts: 493
Joined: Thu Jun 12, 2003 3:04 pm
Location: Dunedin, New Zealand

Return to RPresence

Who is online

Users browsing this forum: No registered users and 1 guest