Cluster sampling question

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

Cluster sampling question

Postby sixtystrat » Tue Dec 15, 2015 11:54 am

I am designing a DNA CMR study for bears based on a 3x3 cluster of traps, with 2 km between traps (i.e., inter-trap distance) and 16 km between clusters (i.e., inter-cluster distance). The population I want to sample is centered on a National Wildlife Refuge but it spills over onto private land, with decreasing bear density as you get further from the refuge. I don't want to have the same sampling intensity in the outlying areas because of the labor involved in setting up traps. My question is whether I can use the 16 km inter-cluster distance in the NWR but gradually increase it (keeping the inter-trap dimensions the same) as I get further and further from the refuge without introducing a bias? I intend to run this scenario in secrdesign but wanted to get some opinions first. Thanks!!!
Joe
sixtystrat
 
Posts: 115
Joined: Wed Jun 01, 2011 9:19 am

Re: Cluster sampling question

Postby murray.efford » Tue Dec 15, 2015 2:59 pm

Hi Joe
If you deliberately apply less sampling effort where there is lower density and do not allow for it in the analysis you will get biased estimates of average density, as you probably intuited. Exactly how you should 'allow for it in the analysis' is not something on which I really want to stick my neck out. It will depend on your goal, and probably requires some sort of density surface model. It's easy enough to stratify the density model for your sampling intensity (presumably assigning more distant habitat to a low-density stratum). If the goal is to estimate an N for the population centered on the refuge then it becomes important to know the density in the outer area, even if it is low, as there is a lot of it - that may take more effort than you would like, as getting precise estimates of low densities still requires a lot of sampling. You might want to impose some biological judgement like 'no bears here' or 'density <10% of refuge'. Sorry not to be very specific here - it's a topic on my 'to-be-investigated' list that has come up in our own bear work.
Murray
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand

Re: Cluster sampling question

Postby sixtystrat » Tue Dec 15, 2015 4:31 pm

Thanks Murray. Yes I was thinking of a density surface model, maybe with distance from the refuge as a covariate. I plan to run some analyses using secrdesign to see how much a stratified versus a uniform trap design might bias the estimates.
sixtystrat
 
Posts: 115
Joined: Wed Jun 01, 2011 9:19 am

Re: Cluster sampling question

Postby sixtystrat » Wed Dec 16, 2015 4:47 pm

I tried to incorporate a non-uniform density distribution in secrdesign but ran into some problems. I get the same message when I run the code in the secrdesign viginette:
xy<-apply(possummask,1,sum)/500
covariates(possummask)[,"D1"]<-2
covariates(possummask)[,"D2"]<-xy-mean(xy)+2.5
scen6<-make.scenarios(g0=0.2, sigma=45, noccasions=5, popindex=1:2)
poplist<-list(list(model2d="IHP", D="D1"), list(model2d="IHP", D="D2"))
sims6<-run.scenarios(5, scen6, traps(possumCH), possummask, pop.args=poplist)

The error is:
Error in runif(N) : invalid arguments
In addition: Warning messages:
1: In rpois(1, lambda = D[1] * area) : NAs produced
2: In rpois(1, lambda = D[1] * bufferarea) : NAs produced
>
Any ideas? Thanks!
Joe
sixtystrat
 
Posts: 115
Joined: Wed Jun 01, 2011 9:19 am

Re: Cluster sampling question

Postby murray.efford » Wed Dec 16, 2015 4:59 pm

Try with
Code: Select all
poplist<-list(list(model2D="IHP", D="D1"), list(model2D="IHP", D="D2"))


Comments: secrdesign is not my favourite software - it's not very forgiving or obvious! You did well to provide a complete working example, so this was easy to troubleshoot.
Murray
murray.efford
 
Posts: 686
Joined: Mon Sep 29, 2008 7:11 pm
Location: Dunedin, New Zealand

Re: Cluster sampling question

Postby sixtystrat » Wed Dec 16, 2015 5:22 pm

Thanks Murray. That did the trick. Happy holidays.
Joe
sixtystrat
 
Posts: 115
Joined: Wed Jun 01, 2011 9:19 am


Return to analysis help

Who is online

Users browsing this forum: No registered users and 11 guests