Approriate method for hand-held VHF data

[johannes Ploderer]

Hi everyone, 

I VHF tracked on foot several individual finches to estimate their range distribution. Our sampling regime was get a fix every 30min from approx. 5am to approx 11am. There are some irregularities in this regime, as it sometimes took us more than 30min to find a finch again. Each day each person was tracking 2-3 birds in parallel. And then we only tracked an individual every 2nd day until we had at least 20 fixes. So on average we tracked an individual for 3-4 days.

I have two questions regarding this dataset

1. Because of this sampling the variogramms and correlorgramms look a bit weird and I am unsure if I can apply the KDEs and AKDEs to them to estimate their occurance distribution. 

2. If there are no worries about using (A)KDEs here is my follow-up question: I already run ctmm.fit() on all individuals and most of them get either IID or IID anisotropic as the best model, but for some OUf is also the best. From what I got from the lecture recordings best practice is to go with the models the function selects, is that also true when the selected models are a KDE (IID) estimate and an AKDE (OUf) estimate or should all individuals be "forced" into a KDE estimate?

Many thanks in advance for your help and all the amazing resources you have put out!

Best, Johannes

[Christen Fleming]

Hi Johannes,

• The visual quality of the variograms doesn't impact the parameter estimations - its just a visual diagnostic.
• You can fix some of that irregularity with the dt option - there is an example in the vignette.
• You can average individual variograms with the mean() function, but yours do all look resident during the observation period.
• You should use the selected model for each individual to account for differences in sampling and movement behavior. If some are IID and some are OUf, that's fine.

Best,

Chris

[johannes Ploderer]

Hi Chris, 

thank you so much!

Best, Johannes