multi-season occupancy: initial occupancy

questions concerning analysis/theory using program PRESENCE

multi-season occupancy: initial occupancy

Postby J.Vanek » Thu Sep 13, 2018 12:32 pm

Hi,

I am just trying to wrap my head around something: initial occupancy. For context, I have 232 sites and ~8 primary sampling periods with 4 potential secondary sampling periods.

When modeling single-species multi-season models (psi, p, gamma, epsilon),

psi is "initial occupancy", correct?

Can anyone clarify exactly what "initial occupancy" is?

1)Is this parameter "occupancy during the 1st primary sampling period only", or "occupancy during the 1st primary sampling period sampled"? What happens if a site was skipped during the first primary sampling period (and coded as missed), and sampled for the first time during the second primary sampling period? Is this site incorporated into the initial occupancy estimate, or is it simply ignored when estimating this parameter?

2) Similarly, are site covariates for "initial occupancy" based solely on occupancy status for the 1st primary sampling period, or for the first primary sampling period with data?

Thanks in advance,
John
J.Vanek
 
Posts: 7
Joined: Sat Jun 04, 2016 9:23 pm

Re: multi-season occupancy: initial occupancy

Postby jhines » Thu Sep 13, 2018 1:17 pm

psi = Probability of occupancy in the 1st primary sampling period, regardless of whether the site was surveyed or not. If a site is never surveyed, you can still estimate the probability of initial occupancy for that site from the sites which were surveyed. If the site was not surveyed in the 1st primary period, it doesn't contribute much information to the likelihood for initial occupancy, but initial occupancy can be estimated for that site using information from the other sites.

Site covariates for psi are used for the estimation of occupancy in the 1st primary period. If a site is not surveyed in the 1st primary period, the probability of initial occupancy for that site can still be estimated from the other sites. For example, if initial occupancy is modeled as a function of elevation ( logit(psi) = B0 + B1*elevation), then sites with survey data allow you to estimate B0 and B1. As long as you have elevation data for sites with missing detection data, you can compute an estimate of initial occupancy for those sites as:

Code: Select all
            exp(B0 + B1*elevation)
  psi = -------------------------------------------
            1 + exp(B0 + B1*elevation)


You can get occupancy estimates for the other primary periods using the equation:

Code: Select all
psi(t+1) = psi(t)*(1-epsilon(t)) + (1-psi(t))*gamma(t)
jhines
 
Posts: 425
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA

Re: multi-season occupancy: initial occupancy

Postby J.Vanek » Thu Sep 13, 2018 1:32 pm

Thank you very much for the clarification! About ~150 of the 232 points were sampled during the first primary occasion, so I wanted to know how much data, if any, was being "wasted."

The data I'm working with was collected in a somewhat complicated way (by necessity as it's over a large area with random sampling not biased towards road access), so I'm wondering how best to organize my primary periods... Most sites were sampled every other year, starting on alternating years, and then another set of "lower priority sites" were sampled every 4 years.

Anyway, thanks again!
J.Vanek
 
Posts: 7
Joined: Sat Jun 04, 2016 9:23 pm

Re: multi-season occupancy: initial occupancy

Postby egc » Thu Sep 13, 2018 3:11 pm

J.Vanek wrote:Thank you very much for the clarification! About ~150 of the 232 points were sampled during the first primary occasion, so I wanted to know how much data, if any, was being "wasted."

The data I'm working with was collected in a somewhat complicated way (by necessity as it's over a large area with random sampling not biased towards road access), so I'm wondering how best to organize my primary periods... Most sites were sampled every other year, starting on alternating years, and then another set of "lower priority sites" were sampled every 4 years.

Anyway, thanks again!


Good to see Jim's reply didn't get categorized as one of the "the absolutely cryptic posts on phidot":

https://www.amazon.com/gp/customer-revi ... dctrvw_srp

Or, perhaps its just the other people trying to help (of their own accord) that are 'absolutely cryptic'. ;-)
egc
Site Admin
 
Posts: 172
Joined: Thu May 15, 2003 3:25 pm


Return to analysis help

Who is online

Users browsing this forum: No registered users and 1 guest

cron