www.phidot.org

As I'm working through secr models, I'm struggling to understand g0 and sigma. My understanding is the g0 is detectibility and sigma is a movement parameter.

Code: Select all

predict(sectmodel)

I get real estimates for density, g0, and sigma.

Code: Select all

species   year   G0   sigma
PETR   2014   0.18920679   31.68196
REME   2004   0.12657903   4.150552


In this case how do interpret g0 and sigma? My understanding is that g0 is the mean detectibility per individual? Or for the species? Also sima for PETR would be a mean distance moved of 32 meters, while REME only moved on average 4 meters?

Thank you.

Hi
In SECR models 'detectability' is modelled as a (typically halfnormal) function of distance, so the parameters g0 and sigma that together define that function replace 'p' in non-spatial models. 'sigma' on its own can be interpreted as an index of home range size, but not directly as 'average distance moved': in some idealized conditions it may equal the sigma of a circular bivariate normal home range model.

In your case, if the sigma estimates are precise then REME would seem to have much smaller home ranges than PETR. The estimates of g0 are just different, and not very different - they should not be interpreted directly as they are meaningful only in connection to a particular sigma.
Murray

Thank you Murray.

So detectibility (g0) in an secr framework does not really have biological significance?

edit-When I plot the secr model (plot(secrmodel)) it shows the relationship between distance (sigma) and detection probability (g0) with some different functions (halfnormal, exponential). I took a class on occupancy and the professor discussed detectibility as a "nuisance" function. Is that a fair way to characterize g0?

sigma is related to home range size but how should I interpret sigma? As I understand, sigma is based off movement of marked individuals. Is there a direct relationship between sigma and home range or it more of general relationship? Bigger sigma, bigger home range?

Lots of questions that I don't really expect awnsers. I'm mostly trying to get a better understanding of the secr models.

Bryan