Segment movement data into pre- and post- home range establishment using variogram

111 views
Skip to first unread message

Tori Mezebish

unread,
May 31, 2023, 11:15:45 AM5/31/23
to ctmm R user group
Hello,

I am working on an analysis of wintering waterfowl and hoping to characterize habitat associations before range residency is established compared to after range residency is established. Initially I thought I could identify the point on the x-axis of the variogram (time-lag) at which the estimated variogram reaches assymptote (see e.g. pic). Then I would segment the location data at that identified time lag. So for the individual whose variogram is below, I would segement the data 1 day into the sampling period - the first 1 day would be "pre home range establishment/dispersal" locations and the remainder of the dataset would be "home ranging" locations.

Is this an appropriate approach? If not, is there another valid means of segmenting data based upon when range residency is established?

Thanks!

Tori
eg_variogram.png

Christen Fleming

unread,
May 31, 2023, 3:35:08 PM5/31/23
to ctmm R user group
Hi Tori,

The x-axis on the variogram is relative time (time lags between data points), where as segmentation happens at absolute times. Sometimes you can see resident behaviors up to a certain timescale in the variogram, which would tell you how long a segment or window might be, but it can't tell you where to place the segment.

This variogram appears range resident, but with either a daily periodicity or duty cycling or something. If there is a day or two of initial dispersal, then it's not readily visible in the variogram. I would suggest plotting the locations colored by time to see that.

I think people mostly do this kind of segmentation by hand or work out some species-specific thresholds on statistics like daily displacement, etc..

I'm actually right in the middle of coding a segmentation fitting method that would estimate the most likely change point, assuming that there is one. I expect an early version to be out this summer.

Best,
Chris

Tori Mezebish

unread,
Jun 1, 2023, 9:21:24 AM6/1/23
to ctmm R user group
Hi Chris,

Thanks so much for your response! Exciting to hear about the path segmentation method - looking forward to playing around with that once it's finished.

I'll pursue the other potential segmentation methods you suggested. One quick follow up to make sure I'm thinking about this correctly... The majority of ducks in my sample displayed similar variograms to the one above (reach asymptote relatively rapidly and show daily periodicity related to a combination of foraging vs roosting behavior and transmitter program). Am I correct in thinking that a rapid achievement of asymptote provides support for the argument that there is a relaitvely short prospecting/dispersal period before individuals establish a home range? So home ranges are established quickly when an individual arrives to the wintering area and, therefore, data segmentation into pre- and post- range residency establishment is not really valid? (with additional support from other metrics such as daily dispersal and locality over time).

Much appreciated!

Tori

Christen Fleming

unread,
Jun 3, 2023, 2:36:36 PM6/3/23
to ctmm R user group
Hi Tori,

It supports either a small or short dispersal. The tracks themselves might be more informative.
I would say possibly unnecessary rather than invalid.

Best,
Chris

Tori Mezebish

unread,
Jun 5, 2023, 9:17:41 AM6/5/23
to ctmm R user group
Got it, thanks so much!

Tori

Reply all
Reply to author
Forward
0 new messages