Suitable trap layouts

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

Suitable trap layouts

Postby howeer » Wed May 12, 2010 4:47 pm

Hi,
At some point, I got it in my head that for the purposes of SECR (assume single session analysis for now), traps should not be isolated in space by more than an home range diameter, because if they were, animals could occupy the area between traps but not be available for capture at any trap. However, I couldn't find refrence to any specific assumption of SECR that would be violated by such a "gap" in a trap array in papers by Efford, Borchers, Dawson, and colleauges, and Royle et al. (2009) specified that secr models "deal with the problem of holes in the trapping array".

Suppose I have spatially explicit capture data from 2 simultaneously-sampled arrays of traps, with trap spacing set to < an average home range radius within arrays, but with > double the average home range diameter separating arrays (assume the entire landscape is homogeneous, suitable habitat, and that animal density does not differ within the greater study area). Numbers of captures and/or recaptures on either of the arrays are only marginally sufficient or insufficient for SECR analysis, but pooled data from both arrays are adequate. Would it be appropriate to estimate density from these data in a single-session SECR analysis?

It occurs to me that exporting an intgration mesh while limiting it to points within a specified distance from traps (say double the average home range diameter) would constrain the points in the mesh to be within the area where home range centers of animals at risk of capture occur, but that a default, rectangular mesh still might not violate any assumptions of the method (though it might slow computation)?

Thanks!
Eric
howeer
 
Posts: 39
Joined: Wed Jun 21, 2006 10:49 am

Re: Suitable trap layouts

Postby murray.efford » Wed May 12, 2010 6:42 pm

Hi Eric
I don't think you have anything to worry about. SECR copes fine with scattered clusters of detectors, and this is one of its major attractions as it lends itself to flexible sampling of regions with randomly or systematically placed clusters. You could look at the early example in Efford et al. (2005) Wildlife Society Bulletin 33: 731-738 which appears now as the 'possum' dataset in secr:

library(secr)
data(possum)
plot (traps(possumCH))
plot (make.mask(traps(possumCH), type='trapbuffer', buffer=100), add=T)

The pooled data analysis is close to the mean of the separate analyses.

You may be thinking of the criterion that detectors within a cluster should not be spaced more than one home range apart: obviously if detectors were that isolated there is no information in the data on the spatial scale of detection.

One thing to be aware of is that a mask constructed with the defaults in make.mask has a fixed number of pixels nx in the x-direction. If clusters of detectors are widely spaced you should increase nx to maintain spatial precision (or set the spacing in metres).

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

Re: Suitable trap layouts

Postby howeer » Thu May 13, 2010 9:25 am

Thanks very much for the clarification.
In some of our sampling units, traps were split into 2 clusters, so I should be fine there. In others, most of the traps were in one cluster, with 2 or 3 traps in an isolated "cluster"... I may still drop those traps, especially if they don't contribute recaptures at different traps, or at least recaptures.

I've manipulated the spacing and x-y distribution of of mesh points using DENSITY (relatively easy to export multiple meshes of user defined extent from a single shapefile of large extent), but will experiment with make.mask.

Thanks again,
Eric
howeer
 
Posts: 39
Joined: Wed Jun 21, 2006 10:49 am


Return to analysis help

Who is online

Users browsing this forum: No registered users and 12 guests