aKDE or HMM for movement with clear phases

[Dibyendu Biswas]

Hi Community!

I have estimated the home ranges of resident old-world vultures for my study area in India using aKDE. But recently I have received a comment that this is not appropriate for such species which may have clear phases of movement regarding foraging or roosting etc. Therefore, I was suggested to use Hidden Markov Model instead.

I need your suggestion in this regard, whether aKDE is sufficient or not.

Thanks in advance!

Sincerely Yours,

Dibyendu

[Christen Fleming]

Hi Dibyendu,

If you have substantially different movement behaviors, then you should segment the data prior to fitting a stationary model in ctmm. A Hidden Markov Model is one way to segment the data, but it is not a substitute for AKDE.

Best

Chris

[Dibyendu Biswas]

HI Christen,

I tried HMM and looked for other options too. Since I am working on the data retrieved from the Movebank I found that the software Firetail (Free version) gives the option to annotate the data as flying, resting etc. based on the 3D acceleration data. Would it be appropriate to first annotate the data and categorize it in different movement phases and then fit the different phase data into the ctmm?

Thanks and regards,

Dibyendu Biswas

Project Fellow

Landscape Level Planning and Management

Wildlife Institute of India, Dehradun

Contact: +91-7739431561 +91-8709642586

SAVE THE ENVIRONMENT

--
You received this message because you are subscribed to the Google Groups "ctmm R user group" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ctmm-user+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ctmm-user/3022c63f-f055-401a-8055-31762411ad0bn%40googlegroups.com.

[Christen Fleming]

Hi Dibyendu,

For the flying data that should work. For the resting data, you would have to be a little more careful, because every location within one resting period is the same.

Best,

Chris

[Dibyendu Biswas]

Hi Chris,

Is it possible to get the different states' data (momentuHMMData class), estimated from "momentuHMM", in the form of CSV/SHP/Raster layer? So that different segments can be modelled in CTMM? I tried write.raster & write.csv function. It didn't work.

Thanks and regards,

Dibyendu

[Christen Fleming]

Hi Dibyendu,

I'm not familiar with that object class, but you would want to get it back into Movebank format to get into ctmm. Either that or get the state at each time and cbind it to the original data.

Best,

Chris

[Genevieve Finerty]

Hi Chris, in what ways should we be careful when working with resting data, just make sure that you really have segmented it correctly? Would averaging across movement phases in the case of a range distribution really matter too much? I had kind of gotten the impression this mattered more for use cases that rely more on the tau[v] params?

[Christen Fleming]

Hi Genevieve,

If you have resting data in different locations, then the time in between those segments needs to be expanded so that the rest speed could have explained those transitions.

With future behavioral switching models, this will not be necessary. But when fitting one stationary model at a time, it needs to be able to explain all of the segmented data.

Best,

Chris

[Genevieve Finerty]

Hi Chris, 

That makes a lot of sense, and I hadn't thought about that. I think that explains some of the issues I've hit previously in path reconstruction with segmented data (day/rest time vs night/move time). Do you have an example of how to expand the time in any of the webinar/workshop videos, or one of the existing ctmm papers?

Gen

To view this discussion on the web visit https://groups.google.com/d/msgid/ctmm-user/bad06881-6301-486a-ba53-38345740e4a8n%40googlegroups.com.

[Christen Fleming]

Hi Gen,

If DATA is the telemetry object and DATA$segment is the segment number and PERIOD is a long period of time much longer than any of the rest timescales and also long enough for the rest movement to explain the journeys between rest sites, then you can just do:

DATA$t <- DATA$t + PERIOD*DATA$segment

Best,

Chris