how to define species observation from non-free-ranging animals

[sarahcd]

Tracking data often represent deployments from homing and navigation experiments that should not be assumed to be movements of free-ranging animals or used to interpret species distributions. Is there a clear way to identify this in DwC or on GBIF?

[timrobertson100]

Thanks @sarahcd - not on GBIF today, no. Will give it some thought though...

[peterdesmet]

Answer by @albenson-usgs:

AB: This is a good question and I don’t have a definitive answer. I would say this could go in extended measurement or fact but I’m wondering if others have an opinion.

[peterdesmet]

Just like outliers (#1) it would be neat to identify this at the occurrence level, so aggregators and users can easily filter those. Ideas (but probably not very good ones):

• basisOfRecord: LivingSpecimen
• establishmentMeans: managed

@tucotuco, I'd like to have your opinion on this.

[tucotuco]

@peterdesmet Your proposal seems quite reasonable. Those are the values I would expect for a zoo animal, for example.

[timrobertson100]

Seems like the best option to me too.

[peterdesmet]

@timrobertson100 @tucotuco Just to be clear, the proposal to use LivingSpecimen, to use establishmentMeans: managed or both?

The definition for establishmentMeans: managed is:

Subclass of introduced: The organism maintains its presence through intentional cultivation or husbandry.

Imo, using establishmentMeans for occurrences has had mixed results and "subclass of introduced" might not be what people expect for all use cases (e.g. tracked wolf in captivity close to national park where it naturally occurs).

---

LivingSpecimen seems like a better (and easier) way to label those occurrences (i.e. known category to exclude when looking for natural occurrences) but those records are also MachineObservations, so how do you indicate that? For reference, this is how we propose to populate basisOfRecord for tracking data:

Kind of event	free range	in captivity
capture	HumanObservation	HumanObs or LivingSpecimen ?
tagging & release	HumanObservation	HumanObs or LivingSpecimen ?
position recorded by device	MachineObservation (lots of them)	MachineObs or LivingSpecimen (lots of them) ?
recapture	HumanObservation	HumanObs or LivingSpecimen ?

[tucotuco]

I see better now where you are going with this. I would keep basisOfRecord completely independent of the establishmentMeans. Observations should be used for records that have no physical evidence - so everything on your list would be Observations for basisOfRecord, unless a sample was taken, or unless it was the "end of the line" (specimen collected). The rest sounds like the vocabularies being discussed are insufficient. I think I would focus energy there to see if you can define values that differentiate the cases that are important.

[peterdesmet]

Great! That's how I see it too: these are all observations, not specimens.

And given the correct controlled vocabulary value (e.g. captive) would establishmentMeans be the correct DwC term to convey that information or are you leaning towards another or new term?

[tucotuco]

The intent of establishmentMeans was, "Why was the thing there?", so that, to me, is the right term. The bothersome part of the definition is "established". Is an individual involved in a homing experiment "established"? If we can get beyond that, I think this is the right term. I think we can get beyond that with a modified definition for establishmentMeans, examples, and a solid vocabulary that addressed the use cases you are facing.

[peterdesmet]

@tucotuco: 👌

@timrobertson/@pieterprovoost, given a new term in the controlled vocab, would GBIF/OBIS be fine in processing this field to detect unnaturally occurring occurrences?