Variogram: measures for meeting asymptote and range-residency assumption

[Luke Emerson]

Hi Chris,

I have produced variograms for puma GPS data and have a couple of questions on how to meet range-residency assumption for calculating aKDEs.

1. Is it appropriate to remove locations that are considered to be part of dispersal events and then just estimate home range from the later locations where it appears that residency has been established? 

For example, the attached plots for Francine indicate eventual range-residency in the variogram (Francine.png), but a clear dispersal event at the start of the monitoring period (Francine_plot.png).

2. If an asymptote is not observed in the variogram and the line is continually trending upwards because it appears an individual's home range has shifted over time, is it appropriate to truncate the data and estimate a home range from just a portion of the data/period of time where residency appears to have been maintained? Particularly when you have hundreds to thousands of GPS fixes over hundreds of days, and hence volume of data or monitoring length is not the issue? 

For example, Bramble.png, variogram doesn't reach asymptote and keeps tracking upwards but has nearly 11000 locations and appears to have established range (Bramble_plot) and was monitored for 850 days. 1.5-2 yrs old when captured so may have been dispersing. If I cut off first 12 months (first 4500 locations) of data, this flattens the data. Is this an appropriate approach? 

Charlotte variogram appears to reach asymptote then keeps going up, possibly due to home range shift westward (Charlotte_plot). 6 yrs old at capture with 13000 data points over 1442 days. Should I leave this data as is, or truncate?

Hoko and Katniss both exhibited variograms that continue to trend upwards and both appear to have possibly shifted home range over time (Hoko_plot and Katniss_plot). Approx 3000 locations and 300 days for each. Would you recommend truncating data, or not estimating home ranges under these circumstances?

3. Should single exploratory forays (or what looks like that) be removed, or should they be kept and considered as a part of the potential home range?

For example, Junior

Regards,

Luke :)

[Christen Fleming]

Hi Luke,

1. Yes, the dispersal range and home range would be two different distributions.
2. Yes.
3. To some degree, that depends on what distribution you are aiming to estimate and how atypical the foray is relative to the fitted movement model. I'd be inclined to keep as much as I could with a flat variogram.
   

Best,

Chris

[Luke Emerson]

Thanks for the advice Chris, much appreciated!

Cheers,

Luke