mask.check : problems with LLonly=FALSE

questions concerning anlysis/theory using program DENSITY and R package secr. Focus on spatially-explicit analysis.

mask.check : problems with LLonly=FALSE

Postby gpugesek » Thu Oct 29, 2015 2:12 pm

Hi,

I am working with a lab group analyzing 4 years of mark recapture data from a population of male butterflies (Becky Fuda is a member of this group). We are using a trap object of a single polygon. I am currently in the process of determining the buffer size of our mask object. I am trying to use mask.check in order to evaluate the effect of varying buffer width on the likelihood estimates of secr.fit. I have identified the point the at which the likelihoods become stabilized using LLonly=TRUE. However, I am unable to use LLonly=FALSE in order to create estimates. When using LLonly=FALSE, I receive this error message "Error in -newfit$fit$minimum : invalid argument to unary operator". I was wondering if you have any insights to this problem.

Thanks,
Genevieve
gpugesek
 
Posts: 9
Joined: Thu Oct 29, 2015 10:32 am

Re: mask.check : problems with LLonly=FALSE

Postby murray.efford » Thu Oct 29, 2015 8:11 pm

Hi
It would help if we could see the code you are using. I would suggest leaving mask.check aside. Instead try fitting a model with secr.fit and checking it with esa.plot. Or just use a buffer 4 x RPSV(CH, CC = TRUE), where CH is your capthist.
Murray
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand

Re: mask.check : problems with LLonly=FALSE

Postby gpugesek » Fri Nov 06, 2015 11:28 am

Thanks so much Murray!

I decided to use a buffer 4 X RPSV (CH, CC=T), as it seemed like the most straight forward approach.

Here is a copy of the code I was using, in which null2012 is the null model for 2012 (I am still unsure why the code LLonly = F would not run).

mask.check(null2012, buffers = (c(0, 10, 25, 50, 75, 100, 150, 200, 250)), LLonly = T)
mask.check(null2012, buffers = (c(0, 10, 25, 50, 75, 100, 150, 200, 250)), LLonly = F)

-Genevieve
gpugesek
 
Posts: 9
Joined: Thu Oct 29, 2015 10:32 am

Re: mask.check : problems with LLonly=FALSE

Postby murray.efford » Fri Nov 06, 2015 4:09 pm

esa.plot should confirm you have a large enough buffer
Murray
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand

Re: mask.check : problems with LLonly=FALSE

Postby N.Coallier » Tue Mar 29, 2016 2:45 pm

This might be a simple question but I don't get how is esa.plot validate your buffer width choice ? Is the curve suppose to be stabilise at the given value ?
Cheers
N.Coallier
 
Posts: 4
Joined: Fri Nov 06, 2015 5:46 pm

Re: mask.check : problems with LLonly=FALSE

Postby murray.efford » Tue Mar 29, 2016 3:03 pm

Yes. By default, esa.plot plots an estimate* of density for a range of buffer widths. Once the buffer is large enough the density estimate stabilizes.

* technically a Horvitz-Thompson-like estimate - the known number detected (n) divided by the inferred effective sampling area (sensu Borchers and Efford 2008) for each buffer width.
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand

Re: mask.check : problems with LLonly=FALSE

Postby N.Coallier » Tue Mar 29, 2016 3:17 pm

Thank you for your quick answer !

Therefore since the curve is stabilise , I would guess a surestimation of the buffer induce less bias in the density estimate than a underestimation ?
N.Coallier
 
Posts: 4
Joined: Fri Nov 06, 2015 5:46 pm

Re: mask.check : problems with LLonly=FALSE

Postby murray.efford » Tue Mar 29, 2016 3:21 pm

If I understand you right, yes. It is OK to have a buffer width that is larger than the minimum.

We usually try to keep the buffer not too large, because processing time increases with the area of the mask. Also, the default mask is discretized with a fixed number of pixels in the x direction (64), and as you increase the width the discretization gets coarser.

Murray
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand


Return to analysis help

Who is online

Users browsing this forum: No registered users and 11 guests

cron