www.phidot.org

Hi Murray and list,

I was talking to David about how to implement Royle and Young's 2008 style analysis in the 'secr' package, where they do a visual search for lizards in a plot and do not have 'trap' locations as such. In its current form, can 'secr' analyse this data type? (I thought not?) David's suggestion was to specify the center of the plot as the trap location and then have a uniform detection function within the plot (of height p, to be estimated), and zero probability of detection outside of the plot.

He thought you had an unreleased version of secr that did strip transects but the detection function decreased with distance from the transect?

Although these idea's are David's, I take full responsibility for any errors in my email

Thanks!
Jo

Hi Jo

There is an undocumented 'polygon' detector type in secr that should do what you want (unless your search area is concave in the east-west direction). Testing and documentation has lagged because I haven't found a good fresh dataset to try it on - maybe you have one?

Here is code for a quick demo

library(secr)
temp1 <- make.poly ()
summary(temp1)
temp2 <- sim.capthist (temp1, noccasion = 4, popn=list(D=20), detectpar=list(g0=0.5, sigma=25), detectfn = 0)
plot(temp2, title=F, vary=T, tracks=T)
secr.fit(temp2, detectfn = 0)

Coordinates for each detection are stored in an attribute of the capthist object that you can extract or assign with the 'xy' method. Check other links from ?xy etc.

Murray