occupancy models, problems with "NaNs"

[joaom...@gmail.com]

Hi all,

I'm conducting single season, single species occupancy models. This model focus on the common genet, with camera trap results from 30 sites during 8 14-day periods (as can be seen in the attached picture). 

However, when trying to model (M1 <- occu(~1 ~1, gg.umf)), I end up getting this warning message: "In sqrt(diag(vcov(obj))) : NaNs produced". The same happens when I backtransform. 

> (M1 <- occu(~1 ~1, gg.umf))

Call:

occu(formula = ~1 ~ 1, data = gg.umf)

Occupancy:

 Estimate  SE   z P(>|z|)

     9.91 NaN NaN     NaN

Detection:

 Estimate   SE     z  P(>|z|)

       -1 0.15 -6.69 2.29e-11

AIC: 268.219 

Warning message:

In sqrt(diag(vcov(obj))) : NaNs produced

> backTransform(M1, type="state")

Backtransformed linear combination(s) of Occupancy estimate(s)

 Estimate  SE LinComb (Intercept)

        1 NaN    9.91           1

Transformation: logistic 

Warning message:

In sqrt(diag(v)) : NaNs produced

At this point I can't disclose the problem and what to do to solve this.

Thanks for any help,

João D.

[Richard Chandler]

Hi João,

The data suggest that psi=1, which means that logit(psi)=Inf. This is the reason for the estimation problem that you've encountered. The bigger issue is that you have very little variation in occupancy to model. If all (or the vast majority) of the sites were occupied, then you can't hope to understand how environmental variables influence occurrence probability.  

Richard

--
You received this message because you are subscribed to the Google Groups "unmarked" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unmarked+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

Richard Chandler

Assistant Professor

Warnell School of Forestry and Natural Resources

University of Georgia

706-542-5815

Lab website

[Matthew Giovanni]

Following up on Richard's comment, you could try modeling abundance (eg w the pcount function) instead of occupancy if all nearly all your sites were occupied.

Matt Giovanni
608-320-9331

--

[joaom...@gmail.com]

Thank you Richard and Matt for your help. In fact, the Genet was one of the species studied with highest naive occupancy (~83%). Modeling abundance might be an interesting approach for the future.

Thanks again,

João D.

[Jessica Gouvea]

Hi everybody,

João I'm with the same problem. I have modeled abundance and had great results. But a doubt came to me: I work with a large-size mammal population (home range about 9km²) in a relative small area (60 km²), so maybe there aren't so many individuals there, considering the home range. So, is this a good model to understand the population?

Thanks