Few Sites, Single Season, but Hatching Season

Skip to first unread message

Court Harding

Aug 10, 2022, 10:17:08 AMAug 10
to unmarked
Hi all,

I'm still new to modeling and want to ensure I am doing best by my data. I have been reviewing various models in this package for several days and want to ensure I am picking the "right" model. I understand that is subject and depends on my objectives.

I am wanting to correlate particular habitat covariates with abundance in a species of iguana. We already know they only occupy a particular habitat (though I did have a few sites unoccupied by the species). Some sites had 2 iguanas, and one had 15+. I have ten sites across an island, and each site was a 100 m transect in which we recorded all iguanas seen. I only sampled over three weeks this summer with different visits to each set of sites due to logistics. Thus, I only have one season of data. However, my sampling occured during hatching season. I did record hatchlings during the surveys.

I feel like I am looking for an open model, but I only have one season of data and not too many sites. I have been considering whether pcount() or pcountopen() is best. I would prefer to model abundance rather than occupancy.

Can anyone point me in the right direction here?

Thanks, Court

Hardin Waddle

Aug 10, 2022, 11:23:30 AMAug 10
to unma...@googlegroups.com
Hi Court-

Did you make at least 2 visits to each of your 10 sites? That (temporal replication) would be a minimum requirement to use these sorts of models.

Also, 10 sites is an extremely small sample size to look for habitat effects. Have you plotted your habitat variables to see how they vary across sites? Are you using categorical or continuous variables?

Not sure why you would think you are looking at an open model. It sounds like your question is around the effect of habitat variables on iguana abundance. Closed in the context of these hierarchical abundance models refers to the population during the sampling season. You would need to have some repeat samples during a period over which you can assume the population was closed to changes in abundance (births, deaths, immigration, emigration). Even the "open" models like pcountopen assume the population is closed across the samples within each primary period.

Although your sample size is small, fitting the data to a model like the N-mixture model in pcount might be informative to you and help with designing future sampling.


*** 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/6483cfee-c930-4c73-8623-18a116f75b53n%40googlegroups.com.

Court Harding

Aug 10, 2022, 11:41:43 AMAug 10
to unmarked
Hi Hardin,

Thanks for your response! I made 4-10 visits to each site. We were under the constraint of time (no further funding to extend the season) and logistics (only so many sites we can visit). Frankly, the island is quite small, so at most we could have doubled the sites if we could also double ourselves. The iguanas are not ubiquitous.

All variables are continuous. Some vary more than others across sites.

I think I misunderstood the closure assumption. I did observe hatchlings at several sites, though I cannot confirm whether they had just emerged or emerged before my surveys.



Aug 11, 2022, 2:27:16 PMAug 11
to unmarked
Hi Court,

You could potentially use an open model and split your surveys. So say if you did 2 surveys pre hatching season and 2+ during or at the end of hatching season. However, open models take a lot of data to run properly and with 10 sites it might be hard to fit a model. You are probably best off using pcount() and modeling Date of survey as a covariate for detection.

On the flip side if you recorded iguanas' distance from transect when observed then you might want to look at distance sampling.


Reply all
Reply to author
0 new messages