Alternative to multi-season occupancy model for count data

[Vatsal Parikh]

Hi all, I have a data for which I want to have different site covariates each season (land cover, population, etc.). I think that this could work well with a multi-season occupancy model (colext in unmarked), but I want to use count values (poisson distribution) rather than presence/absence. Is there a model in unmarked that could be used to achieve this goal? 

[Marc Kery]

Dear Vatsal,

the count-data analogue to the dynamic occupancy model is the Dail-Madsen model, or dynamic Nmix model. This is implemented in unmarked in the pcountOpen function. Can be VERY hard to fit models there when you have covariates. An alternative would be a year-stratified, static model (a.k.a., the "stacked data" approach), i.e., use function pcount() for stacked data and fit a random site effect to account for possible non-independence of the data in different years at the same site. Unless you are particularly interested in the parameters that govern the dynamics, this may be your best choice.

Best regards  --- Marc

---

From: unma...@googlegroups.com <unma...@googlegroups.com> on behalf of Vatsal Parikh <vat...@umich.edu>
Sent: Monday, May 6, 2024 22:33
To: unmarked <unma...@googlegroups.com>
Subject: [unmarked] Alternative to multi-season occupancy model for count data

 

Hi all, I have a data for which I want to have different site covariates each season (land cover, population, etc.). I think that this could work well with a multi-season occupancy model (colext in unmarked), but I want to use count values (poisson distribution) rather than presence/absence. Is there a model in unmarked that could be used to achieve this goal? 

--
*** 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 on the web visit https://groups.google.com/d/msgid/unmarked/95a79ceb-0cad-41b9-bd47-4f4ade55275fn%40googlegroups.com.

[Ryan Williams]

Hi Marc,

I have been trying to figure out how to generate annual abundance estimates using pcount(). I'm working with a large dataset of surveys for a rare turtle species (~700 sites, ~4,000 surveys, 2012-2024). Survey effort across sites and years is not uniform. I have had success with models looking at abundance as related to various site covariates and detection across years. But have been digging through unmarked's documentation, the literature, and this group to figure out how to look at a trend of abundance.

Upon reading about this site/year "stacking" method, I wondered if it would be appropriate to meet my objective. Could I create a y-matrix that looks like the following?

SiteAYear01   survey01 survey02 survey03

SiteAYear02   survey01 survey02 survey03 survey04

SiteAYear03   survey01 survey02 survey03

etc...

Could I then add year as a site covariate to generate estimates of abundance for each year? If that's a reasonable thing to do, and from reading your response to Vatsal, I should also include site as a random effect? If so, does this mean to:

1.) Include site, defined as (using the example above) "SiteA", i.e., the original site ID, or

2.) Include site, defined as (using the example above) "SiteAYear01", i.e., the site as it pertains to the y-matrix

Any input would be greatly appreciated. I am still trying to get my bearings using N-mixture so it is entirely possible I am missing something simple to address my issue.

Thank you!

Ryan