Single season model for 2 years of data, year as covariate

questions concerning analysis/theory using program PRESENCE

Single season model for 2 years of data, year as covariate

Postby Melissa » Mon Nov 27, 2017 5:31 pm

Hello,

I am looking into options for modeling a data set for two years of point count data. I am interested in site covariates that affect occupancy, not necessarily unique estimates of occupancy in each survey year.

Would it be possible to model two years of data with a single season model and include year as a covariate for detection and occupancy, or does this violate the assumption of closure between seasons? Essentially I would be treating 3 survey visits per season times 2 seasons as 6 survey visits in a single season model. A single estimate of occupancy combined over all 6 visits would be acceptable for the purpose of this study, so long as the model does not violate any inherent assumptions that would invalidate the results.

I am considering a single season model as an alternative to an implicit dynamics multi-season model.

Thank you,

Melissa
Melissa
 
Posts: 3
Joined: Tue Nov 21, 2017 2:22 pm

Re: Single season model for 2 years of data, year as covaria

Postby jhines » Mon Nov 27, 2017 8:45 pm

Whether the closure assumption is violated depends on whether the species changes occupancy state over the two years. The ability to detect a change in occupancy over the years will depend on the quantity/quality of the data collected. The safest thing to do would be to start with the multi-season model, allowing occupancy to be different between years, then run a model where occupancy doesn't change (colonization=extinction=0; ie., the single-season model is a special case of the multi-season model). You can then use AIC to let the data indicate whether you need the more complicated model. If occupancy and/or detection is different between years, the single-season model will have a higher (worse) AIC value and give an occupancy estimate between the two yearly estimates.

With the multi-season model, it's possible to build a model where occupancy is affected by covariates, and different between years, but the effect of the covariates is the same for both years. For example, there might be a positive effect of a particular habitat-type near the sites for both years, so you'd have 4 occupancy estimates:

$$\psi_1^{poor habitat}, \psi_1^{good habitat}, \psi_2^{poor habitat}, \psi_2^{good habitat}$$, with the constraint that the difference in occupancy between poor and good habitat is the same for both years ($$\psi_1^{good} - \psi_1^{poor} = \psi_2^{good} - \psi_2^{poor}$$ on the logit scale). This works for a continuous covariate as well (eg.,

$$\psi_1 = \beta_0 + \beta_1*elevation$$\newline
$$\psi_2 = \beta_2 + \beta_1*elevation$$
jhines
 
Posts: 599
Joined: Fri May 16, 2003 9:24 am
Location: Laurel, MD, USA


Return to analysis help

Who is online

Users browsing this forum: No registered users and 8 guests