PKDE area outputs vs traditional KDE methods

[Gary C]

Hi all, 

I was hoping if anyone could advise or sense check suitability of methods for a dataset. 

I have 30 min GPS data from a central place forager seabird species. Tracking is short term (up to a week with discrete trips of 1-2 days typically). 

As expected, variograms showing autocorrelation within trips and cyclical non-rage resident behaviour and low DOF. 

I have been exploring AKDE and PKDEs based on individual trips as the unit instead of the individual animal. 

When comparing outputs with more traditional seabird KDE methods (e.g. apply CRAWL then adehabitat package using href bandwidth) the PKDE outputs were broadly similar but often smaller and more restricted (in contrast to Noonan et al 2019 typically showing larger areas for AKDE). 

From what I have read (including new Alston et al paper) this is perhaps expected given the type of data (not collected over long range resident time period) and some of the furthest trips from the colony are not included in PKDE if apparently 'exploratory'. 

- Are there then any recommendations of sensible ways to filter/format short term central place forager data for use with AKDEs or is it best to focus on occurrence distributions (if question is overlap with areas of interest offshore)?

Happy to provide examples if helpful. 

Many thanks, 

Gary

[Christen Fleming]

Hi Gary,

If you feed one trip at a time through AKDE (before PKDE) then the individual trip foraging areas will be underestimated, because the trip DOFs will be 1-2.

For CPF individuals, generally you want to segment out the central place and work with the foraging trip data. Then you want to have enough representative trips for the AKDEs. With multiple individuals, aggregating those into PKDEs does make sense.

Best,

Chris

[Gary C]

Many thanks for the reply Chris and I appreciate your support on this group and materials available online for ctmm. Incredibly helpful.

In the example I shared (apologies without full details) yes the data are segmented to locations on foraging trips only and no data from the colony are included the AKDE/PKDEs so I was thinking in a more general sense of the application of these methods to central place foragers. The PKDE outputs for some years (aggregating multiple trips and individual within years) are not capturing the observed locations most distant from the colony and wondered if a fundamental issue with using short trips (1-2 days).