HELP! Using variogram to decide if data should be subsetted for home range estimation

[Jamie Bolam]

Hi all,

Please see attached my variogram - as you can see the animal looks like it has a rough asymptote up until month 7 or so, then it decreases a bit. The animal then moved in the past 3 months to a completely new location (see red points on the telemetry plot), which is not too far, but does not overlap at all with the previously occupied area, so the variogram has curved up steeply at the end.

If I need to do weighted AKDE home range/utilisation density estimates, do you recommend I analyse this entire dataset, with the last 3 months of movement included? Or should I subset the dataset to generate AKDEs based on the first 8-9 months? Where would you recommend subsetting it, if you would? Based on this CTMM lecture, which handily has an example of CTMM's power compared to KDE in the first few minutes (https://streaming.uni-konstanz.de/paella/?tx_uknkimstreams_player%5Baction%5D=player&tx_uknkimstreams_player%5Bcode%5D=AniMove_2022-09-16_01&tx_uknkimstreams_player%5Bcontroller%5D=StreamList&tx_uknkimstreams_player%5Btitle%5D=&cHash=c6a6054ec5e75575378f3313371c682e) I hope to use the AKDEs to show the (hopefully) future-proof home range of my study animal including this new foray into a new area. My group has previously just used MCPs and KDEs.

I have a report due in 1.5 weeks and need to include these results. This is a work report (alas no longer a student) so it will have real-life conservation impact - your advice is greatly appreciated!

Cheers!

Jamie

[Christen Fleming]

Hi Jamie,

There is a graduate clock-wise range shifting. You can try to fit one stationary model to the entire dataset and see how well that matches the empirical variogram.  You might also try adding the circle=TRUE argument in your ctmm.guess() object to see if it can capture that circulation.

You can also try ?3? spatial clusters or space-time clusters, estimating those separate ranges, and then mean() them together into one larger range. That would be less sensitive to the non-stationary range shifting, but the calculation would assume that shifting will continue only within this area.

Best,

Chris