Single species, seasonal variation in occupancy

[Gareth Walker]

Good day all,

I was hoping that you could offer some guidance.

I am fitting a single species occupancy model from data collected during a single year between the wet and dry seasons. In total, 20 camera traps were deployed for a period of 30 days each season. In short,  my aim is to evaluate temporal variation in occupancy for 13 individual species relative to three site specific covariates.

My data are as a follows:

Detection (y) <- matrix

Site_covs <- Forage cover (percentage), forage nutritive quality, distance to nearest rivers (m).

Obs_covs <- Habitat type (LRB) and season (wet vs dry)

I have attempted a variety of different approaches however it seems as if I am going around in circles. For example, I have evaluated temporal variation in occupancy on a per season (1) basis using the following code:

m3 <- stan_occu(~Season:Habitat ~Season:Habitat + Season:scale(Grass_cover) + Season:scale(Shrub_cover) + Season:scale(Tree_cover) + Season:scale(Grass_quality) + Season:scale(Shrub_quality) + Season:scale(Tree_quality) + Season:scale(Distance_to_rivers), umf, chains=4, iter=10000) 

Similarly, I have evaluated temporal variation in occupancy relative to habitat (2) using the following code:

m4 <- stan_occu(~Season:Habitat ~Season:Habitat + Habitat:scale(Grass_cover) + Habitat:scale(Shrub_cover) + Habitat:scale(Tree_cover) + Habitat:scale(Grass_quality) + Habitat:scale(Shrub_quality) + Habitat:scale(Tree_quality) + Habitat:scale(Distance_to_rivers), umf, chains=4, iter=10000)

Both approaches suggest that no covariates have an effect on the species occupancy. My question is two part:

1) Will I be able to infer reliable estimates using the above approach?

2) Is it necessary for met to scale my grass, shrub and tree cover variables given that these are estimated percentage estimates on a per site basis? For example: 

 m22 <- stan_occu(~Season:Habitat ~Season:Habitat + Season:Grass_cover + Season:scale(Grass_quality), umf, chains=4, iter=10000)

If I do this, the results suggest a significant effect however I receive a warning about covariates not being standardized (1) and divergent transitions (2).

Any help would be greatly appreciated!

Regards

Gareth

[JOSE DOMINGO INFANTE VARELA]

Hi Gareth,

It seems to me that your sample size (40 sites right?) is too small for three covariates. It's hard to get significant statistical associations if your sample size is small, even if biological effects/differences exist.   

Perhaps someone else in the group has a more elaborate and accurate response.

Best,

José.

---

De: unma...@googlegroups.com <unma...@googlegroups.com> en nombre de Gareth Walker <gareth.wal...@gmail.com>
Enviado: martes, 30 de agosto de 2022 3:37
Para: unmarked <unma...@googlegroups.com>
Asunto: [unmarked] Single species, seasonal variation in occupancy

 

--
*** 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/c561a6d8-70c1-4554-9093-31e740651c6an%40googlegroups.com.

[Gareth Walker]

Good day Jose,

Apologies for the late reply.

Thank you for your input. You were spot on. Too few replicates over too short of a time period.

Thanks,

Gareth

To view this discussion on the web visit https://groups.google.com/d/msgid/unmarked/MN0PR11MB6159A63ACD0DD2E9114CEB60D2799%40MN0PR11MB6159.namprd11.prod.outlook.com.

[JOSE DOMINGO INFANTE VARELA]

Hi Gareth,

I know how hard it is to get samples for ecological studies. Also, models based on 0/1 data are too data hungry. Maybe you should explore temporal variation in activity instead of occupancy by using monthly detections. It can be done using a GLMM with an effort offset (number of operating nights per month and camera) and a site random effect. 

Cheers,

José.

---

De: unma...@googlegroups.com <unma...@googlegroups.com> en nombre de Gareth Walker <gareth.wal...@gmail.com>

Enviado: sábado, 3 de septiembre de 2022 2:22
Para: unma...@googlegroups.com <unma...@googlegroups.com>
Asunto: Re: [unmarked] Single species, seasonal variation in occupancy

 

To view this discussion on the web visit https://groups.google.com/d/msgid/unmarked/CAPZAkGFWqO9K2d7x9aA1ZCuGY-2JUH%2BjspVQmdv%2BRZR0YQMV_w%40mail.gmail.com.