I ave tried with the adehabitaHR package but my boundary does not satisfy the angle criteria between lines and I can't fiddle with that.
Any other solution??
Have you tried T-LoCoH or Brownian Bridges - they should be able to deal with hard boundaries
Thanks for the answer Femke. Nope, haven't tried LoCoh cause I wanted to stay with the standard KDE... I suppose you are referring to the tlocoh and BBMM packages right?
I managed to create KUDs specifying the boundary argument in the kernelUD function for wild dogs in Hluhluwe-iMfolozi Park in SA. This park has hard boundaries with predator proof fencing enclosing the entire protected area. I created a "virtual boundary" using the locate function to define the turning points of this boundary so that each segment was at least 3*h (h=smoothing parameter) and turning angles were all greater than 3*pi so that the virtual boundary was the least tortuous it could be. I have included a map of the park so that you can see what the normal boundary looks like and the tight turning angles in the north, south west and south east that needed a lot of attention. I'm not exactly sure what your study site looks like but how much fiddling with the boundary are you willing to do?
The whole process to create the boundary took a long time and a lot of fiddling to get it exactly right (note here that this virtual boundary did not exactly match the actual boundary). The KUDs created looked appropriate (not a great deal of over smoothing either).
I also created territories using LoCoH (k-LoCoH specifically) which built territories well that incorporated the effect of the hard boundary. However, it seemed to drastically underestimate territory size (when comparing to the literature) so we ultimately stuck with the boundary KUDs (the virtual boundary not being exactly aligned with the actual boundary).
Hope this helps. If you are interested in the code used to create the boundary, I'm happy to share.
Hi David, thanks a lot for your reply! Yeah my boundary looks rather like mine with sharp turning angles (fences). I'm not sure I'm prepared to fiddle with my boundary make it fit the criteria for the adehabitatHR package in terms of both angle and length... But just in case I decided to, I'm definitely interested in the code, thanks a lot!
I think I'm gonna have to use the LoCoH (I'm also concerned with HR size...) and maybe BBMM...
Does it need to be Kernel? What about Local Convex Hull? It may be a satisfactory alternative for your problem?
Just a thought which crossed my mind. Could you clip your UD by the boundaries and then re-scale the remaining values? This might be a lot easier to do than David's approach. I suspect the outcome could be quite different from the approach David took though. In addition to that, I'm not sure if either result would be a good reflection of the area use. In some areas wild dogs really use fences and probably spend more time there than expected, so you would have an area of (very) high use right next to an area of zero use (other side of the fence). The latter you could only really achieve, I think, with clipping.
Thanks for your answers guys.
Gabriele Cozzi, I'm not a big fan of the Hull look (just on the graphical point of view) , hence was looking for a smoother representation like kernels. But I'll definitely look into it as well.
Egil Droge , that sounds like a very good idea to me! I had just clipped (which work for representation but not for actual values) but re-scaling would definitely be a great approach! Lions also used areas along fences quite a lot here. I shall try that.
Kernels can not deal with hard boundaries. Gabriele is right Local Convex Hulls is your candidate for that.
Ok LoCo it is then hey?
Thanks for the reply Dominik Thomas Bauer !
Stéphanie Périquet also now facing this same problem here. I found this paper https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4803229 and the permissible home range method seems a good option for enclosed populations as the LoCoH for sea otters still overlapped with the land. Might be good to chat when I next see you? (LCMAN?)
Hi Sarah Edwards , thanks for that hey! We should absolutely chat about this and so many other things ;-) No idea if I'll go to the LCMAN meetings but I'm sure we'll make a plan!
Resistant kernel with high resistance along barriers. suzy Rostro did this in one of her earlier papera I think following my suggestion. Didn’t we discuss this before?
Can "home ranges" for large mammals really be defined across algorithms? Seasonal, annual, life-long and/or multi-generational ranges? So far I failed to identify "home" ranges empirically for elephant herds, solitary bears or other large mammals. Depends probably on definition of "home". For example, could "home" be considered as synonym for habitat? If interested, please be welcome to answer my pertinent question and provisional answer on this site. So far, the scientists on this site have not proposed a comprehensive definition to distinguish between these related paradigms with the same set of attributes.
- Badges
- Science topic
- Similar topics
- Psychology
- Social Psychology