Interpreting occupancy vs. use in multiple-season models

54 views
Skip to first unread message

Carolina Jankowicz

unread,
Apr 24, 2026, 12:25:48 PM (3 days ago) Apr 24
to unmarked

Hi, how are you?

I have a question regarding the possible interpretations of occupancy/use probability in a multiple-season model.

I have been working with single-season models, and in most cases—due to ecological and design considerations—we have found it preferable to interpret the probability as use rather than occupancy, since it was not clear that the species would be present continuously at what we defined as a site throughout the entire season.

However, I find it difficult to extend this criterion to multiple-season models. At least in the explicit modeling framework, where colonization and extinction are estimated: can I refer to a site as being “used,” and then talk about it being colonized or going extinct? I understand that, in numerical terms, if one interprets the parameter as use and estimates colonization, this would represent the rate or probability that use increases from one season to another. But ecologically, does this make sense?

In the implicit model, this feels less problematic, as I understand that I would simply be interpreting different probabilities of use that a site may have across seasons.

When it comes to designing the sampling scheme—particularly the distance between sites—it felt easier to apply a criterion in the single-season case. Even with some uncertainty about how much the species might move over certain time intervals, I could still find a way to interpret things appropriately based on the criteria used. In this case, with the added confusion about what colonization and extinction would imply under a “use” interpretation, I find it more difficult to interpret things ecologically.

  I would greatly appreciate your guidance on this matter,

Kind regards,

Carolina


Josh Twining

unread,
Apr 24, 2026, 1:34:31 PM (3 days ago) Apr 24
to unma...@googlegroups.com
Hi Carolina,

Many potential approaches and responses to this, but you might be able to navigate/side step - first question would be, what are your objectives / questions of interest with this study/analysis? and if colonization and extinction estimates are not relevant / central to them, you may consider just fitting a stacked single-season model (e.g., each site x year combination is considered a unique site, and you fit year as covariate on both psi and p to account for the artificially increased precision in parameter estimates). This would navigate your challenges in interpretation of the transition parameters while interpreting psi as use, it would also save you some statistical power by not estimating additional parameters you are not interested in. 

All the best

Josh

--
*** Three hierarchical modeling email lists ***
(1) unmarked (this list): for questions specific to the R package unmarked
(2) SCR: for design and Bayesian or non-bayesian analysis of spatial capture-recapture
(3) HMecology: for everything else, especially material covered in the books by Royle & Dorazio (2008), Kéry & Schaub (2012), Kéry & Royle (2016, 2021) and Schaub & Kéry (2022)
---
You received this message because you are subscribed to the Google Groups "unmarked" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unmarked+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/unmarked/b200568d-d8eb-4ab0-97e5-79318b77002bn%40googlegroups.com.


--
Joshua P. Twining, PhD,
Assistant Professor,
126 Nash Hall,
Oregon State University,
Oregon, 
97331

Message has been deleted

Carolina Jankowicz

unread,
Apr 25, 2026, 10:48:33 AM (2 days ago) Apr 25
to unmarked

Hi Josh,

Thank you for your response. I am working in a study area where the species was not present at the beginning of the study and colonized the area over the course of the sampling period (a process that took several years). I have sampling sites distributed across this area, and I am interested in understanding how this process unfolded: when did the species appear at each site? Is it possible to identify a spatial pattern in its return (for example, appearing first from the south)? I am not sure whether I will be able to characterize any form of intensity (i.e., few initial records vs. a higher frequency of records at the same site over time). I do not yet have those details fully defined, but broadly speaking, my goal is to characterize this type of process.

So I do expect site states to change over years, and I am interested in that change (as well as in incorporating potential covariates, e.g., distance to other potential sources). Do you think it makes sense to use a multiple-season framework?

Returning to my initial question, given the variability in the species’ movement patterns, I am not sure whether I can define a site as a unit that is either occupied or not, and I also do not think that would be particularly useful. I believe that a “use” interpretation is more appropriate. However, this is where my concerns from the initial message come into play.

Thank you very much for your time,

Carolina  

Quresh Latif

unread,
Apr 25, 2026, 8:19:48 PM (2 days ago) Apr 25
to unma...@googlegroups.com
Modeling dynamics seems perfectly suited to your question, and I don't see any problem with modeling dynamics with your intended interpretation of occupancy. Occupancy for you will be the probability that the species ever occurs at a given site in a given year, or possibly if your species is territorial, the probability of the site intersecting at least one breeding territory. Colonization would then be the probability that the site becomes occupied by the territory or home range of at least one individual where it was not in the previous year. This interpretation still supports inference of changes in species distribution over time, including patterns in how and where the species colonizes and spreads.

The one thing you'll need to watch out for is low detectability bias. Since your detectability includes both observer error and movement of individuals between repeat surveys, detectability will be lower than it would if you were conducting removal sampling (i.e., breaking up a single survey into smaller intervals across which movement is less likely). When detectability is too low, occupancy (and possibly colonization?) estimates may become biased. I would check the literature reporting on simulations for dynamic occupancy models and make sure your detectability is not too low to have confidence in your occupancy and colonization estimates.

Quresh S. Latif
Biometrician
Bird Conservancy of the Rockies
230 Cherry St., Ste. 150, Fort Collins, CO 80521
970-482-1707 (ext. 15)
Connecting people, birds and land




--
Joshua P. Twining, PhD,
Assistant Professor,
126 Nash Hall,
Oregon State University,
Oregon, 
97331

--
*** Three hierarchical modeling email lists ***
(1) unmarked (this list): for questions specific to the R package unmarked
(2) SCR: for design and Bayesian or non-bayesian analysis of spatial capture-recapture
(3) HMecology: for everything else, especially material covered in the books by Royle & Dorazio (2008), Kéry & Schaub (2012), Kéry & Royle (2016, 2021) and Schaub & Kéry (2022)
---
You received this message because you are subscribed to the Google Groups "unmarked" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unmarked+u...@googlegroups.com.

Marc Kery

unread,
Apr 26, 2026, 4:07:04 AM (yesterday) Apr 26
to unma...@googlegroups.com
Dear Carolina,

given your interest in the colonization of your study area and factors driving it such as distance to source sites, you might perhaps want to consider models that make colonization a function of a measure of connectivity with occupied sites. E.g., like this paper: https://esajournals.onlinelibrary.wiley.com/doi/10.1890/15-0416.1. Thing is you will have to go to Bayesian BUGS software for it.

We have something on these models in the AHM2 book (9.6 MECHANISTIC, OR DYNAMIC, MODELS OF SPATIAL AUTOCORRELATION) and you can download the code for free from the website.

Best regards  --- Marc

 


From: 'Quresh Latif' via unmarked <unma...@googlegroups.com>
Sent: Sunday, April 26, 2026 02:19
To: unma...@googlegroups.com <unma...@googlegroups.com>
Subject: Re: [unmarked] Interpreting occupancy vs. use in multiple-season models
 
Reply all
Reply to author
Forward
0 new messages