Distance sampling in unmarked: Violations of detection ~ distance assumption for some bird species

[Gabri Miret]

Hi all,

I'm working with transects data for birds and fitting distance sampling models using the gdistsamp function in the unmarked R package. My distance bins are: 0–25 m, 25–100 m and 100–1000 m (as the surveys are done following these bins too). 

For most species, things behave as expected — higher counts in closer bins and fewer at longer distances. But for a few species, I’m observing the opposite: most detections are in the farthest bin (100–1000 m), and very few in the closest one. This seems to violate the assumption that detection probability declines with distance, which is fundamental to distance sampling. Probably, these happen due to issues with species-specific detectability (e.g., loud calls detectable far away).

My questions are:

1. What are the best practices in this case — is it valid to fit distance sampling models at all for these species?

2. Is there any workaround in unmarked::gdistsamp to handle this violation (e.g., alternate key functions, model truncation, or bin adjustments)?

3. Would you recommend switching to a different model type for these species?

4. Could this reflect a problem with the distance bins themselves (e.g., too wide/few)?

I’d appreciate any advice or references on handling this kind of issue. Thanks in advance!

Best,

Gabri

[Jeffrey Royle]

hi Gabri,

 From what I understand here, your 3rd distance bin spans 100 m - 1000m and as such is much wider than the other two. Therefore you might expect more detections in that bin simply because the expected number of birds in that bin is 9x the other two bins combined.  Therefore I don't think you have a problem here.

regards

andy

--
*** 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/d8a3ff3b-d1a4-45ae-8bcc-7b51b06795e9n%40googlegroups.com.

[Marc Kery]

Dear Gaby,

I think your data may still adhere to the model's assumptions, since your final bin is extremely wide. If p is not very small, then I think it's possible to have more detections in that bin than closer-by. I think you should make a parametric bootstrap GoF to test this assumption, rather than just visually checking the raw data.

Best regards  --- Marc

---

From: unma...@googlegroups.com <unma...@googlegroups.com> on behalf of Gabri Miret <gabri...@gmail.com>
Sent: Thursday, July 3, 2025 14:23
To: unmarked <unma...@googlegroups.com>
Subject: [unmarked] Distance sampling in unmarked: Violations of detection ~ distance assumption for some bird species

 

--

[Marc Kery]

you beat me ;)

---

From: unma...@googlegroups.com <unma...@googlegroups.com> on behalf of Jeffrey Royle <jar...@gmail.com>
Sent: Thursday, July 3, 2025 14:31
To: unma...@googlegroups.com <unma...@googlegroups.com>
Subject: Re: [unmarked] Distance sampling in unmarked: Violations of detection ~ distance assumption for some bird species

 

To view this discussion visit https://groups.google.com/d/msgid/unmarked/CAE%2BwgF1WmkfFO7Nq_ixqH2WkvBfVAtBZf_HemtO3g5m1kcBJEw%40mail.gmail.com.

[Gabri Miret]

Dear Andy and Marc,

Thank you very much for the prompt answer, that makes sense! I hadn’t fully accounted for how much larger the area is in the last distance bin, so I guess it can be ok. I’ll keep this in mind moving forward.

Thanks again!

Best,

Gabri

El dia dijous, 3 de juliol del 2025 a les 14:32:44 UTC+2, Kery Marc va escriure: