Partial Migratory behaviour and variograms

[Licia Calabrese]

Dear Dr. Fleming,

 

I have a doubt in interpreting two ctmm variograms built on deers telemetry data.

For the first one (stationary individual, Fig1) I have 14 months of data and judging from the map, the deer tends to stay in the same area (about 5 km long and 1.7 km width). However it seems that the variogram in Fig 1 keeps on growing without finding athe horizontal asymptote.  This, from what I read, means that the deer is not stable in the area. However the deer occupies the same area all the time. Can I still model a stable HR even if the variogram behave like that?

 

The second deer (3 years of data, Fig2a and b) shows a migratory behaviour that is repeated every year as it moves for 4 months in a different area. I was thinking to cut the data, however, I would have to remove the data where the deer exhibits an exploratory behaviour (it stays in the “migration area” for a few days and then it goes back to the main area for another few days before moving in the migration area for 4 months). That would be a pity and then I would end up with 2 HRs per year rather than a general one. How would you obtain a general HR using all the data? I read the periodogram vignette and I was wondering if there was any development on the coding for the home range calculation in cases of periodic/migratory behaviour.  

 

Thank you very much for your help

 

Licia

[Christen Fleming]

Hi Licia,

For deer 1, if I had to guess, they are probably exhibiting multi-scale behavior where they use a small transient range that moves around within the larger 5 x 1.7 km^2 range. I've seen this with deer and elephants. The multi-scale models required are not yet in the package, and probably will not be for a couple of years. Alternatively, there could be some issue with data quality (such as bursts of high frequency sampling) that requires an error model or thinning the data down to where the error doesn't matter.

For deer 2, the ideal analysis is not yet implemented, but what I would do to approximate it is to segment the data in to 3 behaviors (summer, winter, migration), and then for each behavior (and using all years if there is fidelity) estimate the ranging area, and then average the 3 ranges together with the mean() function. I need to update the CIs for that calculation, but the point estimate should be good.

Best,

Chris

[Licia Calabrese]

Dear Chris,

Thank you very much fo your quick reply. 

For deer 1 data quality is not too bad I think, I have a few gaps of 1 day every 1-2 months and a big gap of 5 days. The deer was probably in some area with poor GPS coverage. It could be a multiscale bahaviour as you said I think. In such case, can I still use the ctmm function ?or would you suggest another method?

For deer 2, I did what you suggested and the variogram for area 1 (split1variogram_full, calculated using only the area 1 locations pooling all years together) seems OK but in area 2 (split2variogram_full) there are temporal gaps, occurring because I've truncated te data. Are these gaps affecting the winter HR estimation? Can I use this variogram to calculate the 3-years winter HR? I'm a bit concerned that the model would give more "weight" or importance to the locations at the beginning and at the end of the wintering period. I'm not sure if it's clear what I'm trying to say... Does the mean() function average only the areas or it also acts on the perimeter of them?

Thanks a lot again!

Best Regards

Licia

--
You received this message because you are subscribed to a topic in the Google Groups "ctmm R user group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ctmm-user/gd2OwJdB_6s/unsubscribe.
To unsubscribe from this group and all its topics, send an email to ctmm-user+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ctmm-user/0e85e046-0c9d-4b09-8c94-cd72a6806e59%40googlegroups.com.

--

Licia Calabrese, PhD
Conservation Biologist

[Christen Fleming]

Hi Licia,

Gaps make the variogram ugly, but are not generally a problem otherwise. What can throw the model off (when error is not included) is short bursts or tiny timesteps in the data.

If I had to use data like deer 1, I would look to segment the data (if there are discrete shifts) or do a sliding window analysis (if there is continuous shifting) and aggregate the transient ranges together.

The mean() function averages the density functions, but corresponding CIs are not currently estimated with any rigor, so let me know if you need that implemented beforehand.

Best,

Chris

To unsubscribe from this group and all its topics, send an email to ctmm-user+unsubscribe@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/ctmm-user/0e85e046-0c9d-4b09-8c94-cd72a6806e59%40googlegroups.com.

[Licia Calabrese]

Dear Chris,

Understood.

Thanks a lot for everything, I'll let you know if I will need to use the mean() function. 

Best Regards,

Licia

To unsubscribe from this group and all its topics, send an email to ctmm-user+...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/ctmm-user/0e85e046-0c9d-4b09-8c94-cd72a6806e59%40googlegroups.com.

--

Licia Calabrese, PhD
Conservation Biologist

--
You received this message because you are subscribed to a topic in the Google Groups "ctmm R user group" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/ctmm-user/gd2OwJdB_6s/unsubscribe.

To unsubscribe from this group and all its topics, send an email to ctmm-user+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/ctmm-user/dea82298-3ae2-4128-a8a2-57e83e55fe75%40googlegroups.com.